CN117216481A - Remote monitoring method and system for electric appliance - Google Patents
Remote monitoring method and system for electric appliance Download PDFInfo
- Publication number
- CN117216481A CN117216481A CN202311283691.6A CN202311283691A CN117216481A CN 117216481 A CN117216481 A CN 117216481A CN 202311283691 A CN202311283691 A CN 202311283691A CN 117216481 A CN117216481 A CN 117216481A
- Authority
- CN
- China
- Prior art keywords
- data
- target
- fault information
- remote monitoring
- parameter set
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 74
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000012216 screening Methods 0.000 claims abstract description 80
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 39
- 238000012545 processing Methods 0.000 claims abstract description 39
- 238000010219 correlation analysis Methods 0.000 claims abstract description 32
- 239000011159 matrix material Substances 0.000 claims description 47
- 238000009826 distribution Methods 0.000 claims description 24
- 238000004458 analytical method Methods 0.000 claims description 22
- 238000012549 training Methods 0.000 claims description 20
- 238000013506 data mapping Methods 0.000 claims description 16
- 230000004927 fusion Effects 0.000 claims description 15
- 238000003860 storage Methods 0.000 claims description 15
- 238000010276 construction Methods 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 13
- 238000011161 development Methods 0.000 claims description 13
- 238000012806 monitoring device Methods 0.000 claims description 13
- 238000000605 extraction Methods 0.000 claims description 11
- 238000010586 diagram Methods 0.000 claims description 9
- 238000009499 grossing Methods 0.000 claims description 7
- 230000010354 integration Effects 0.000 claims description 7
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 230000000875 corresponding effect Effects 0.000 description 21
- 238000012423 maintenance Methods 0.000 description 14
- 230000008569 process Effects 0.000 description 10
- 238000001514 detection method Methods 0.000 description 9
- 239000000284 extract Substances 0.000 description 7
- 238000001914 filtration Methods 0.000 description 6
- 238000004519 manufacturing process Methods 0.000 description 6
- 238000000638 solvent extraction Methods 0.000 description 6
- 238000010801 machine learning Methods 0.000 description 5
- 230000002159 abnormal effect Effects 0.000 description 4
- 230000002596 correlated effect Effects 0.000 description 4
- 238000010606 normalization Methods 0.000 description 4
- 238000007405 data analysis Methods 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- 238000005192 partition Methods 0.000 description 3
- 230000000717 retained effect Effects 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 238000012935 Averaging Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 238000010200 validation analysis Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013075 data extraction Methods 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000003862 health status Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000013450 outlier detection Methods 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Landscapes
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention relates to the technical field of data processing, and discloses a remote monitoring method and a remote monitoring system for an electric appliance, which are used for improving the accuracy of remote monitoring for the electric appliance. Comprising the following steps: inputting the target time sequence feature set into a target SF-ELM model to predict fault information, and obtaining initial predicted fault information; performing spatial correlation analysis based on the initial prediction fault information to obtain spatial correlation data, and constructing topological relations of a plurality of sensors based on the spatial correlation data to obtain target topological relations; according to the target topological relation, carrying out data screening on the initial prediction fault information through a preset mutual information entropy screening algorithm to obtain screening fault information; and carrying out data reconstruction on the initial predicted fault information based on the screening fault information to obtain target predicted fault information, and transmitting the target predicted fault information to a preset data processing terminal to carry out remote monitoring on the target electric appliance.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to a remote monitoring method and a remote monitoring system for an electric appliance.
Background
Electrical appliances play an important role in modern society, covering various fields from home appliances to industrial appliances. As the complexity and number of electrical devices continue to increase, their stability and reliability become critical. Therefore, the real-time monitoring and fault detection of electrical equipment become an important research field.
Traditional fault detection methods often rely on offline data analysis, and real-time monitoring and prediction are difficult to achieve. Meanwhile, the collection and processing of a large amount of data from multiple sensors involves complex tasks such as data cleaning, denoising, and feature extraction. These steps require a lot of time and resources and in some cases the data quality is affected by noise, i.e. the accuracy of the existing scheme is low.
Disclosure of Invention
The invention provides a remote monitoring method and a remote monitoring system for an electric appliance, which are used for improving the accuracy of remote monitoring of the electric appliance.
The first aspect of the present invention provides a remote monitoring method for an electric appliance, the remote monitoring method for an electric appliance comprising:
acquiring real-time parameters of the electric appliance through a plurality of sensors installed in the target electric appliance to obtain a real-time parameter set;
Extracting time sequence characteristics of the real-time parameter set to obtain a target time sequence characteristic set;
acquiring a historical parameter set of the target electrical appliance, performing data division on the historical parameter set to obtain a plurality of division data sets, and performing model training on a preset initial SF-ELM model through the plurality of division data sets to obtain a target SF-ELM model;
inputting the target time sequence feature set into the target SF-ELM model to predict fault information, and obtaining initial predicted fault information;
performing spatial correlation analysis on a plurality of sensors based on the initial prediction fault information to obtain spatial correlation data, and performing topology relation construction on the plurality of sensors based on the spatial correlation data to obtain a target topology relation;
according to the target topological relation, carrying out data screening on the initial prediction fault information through a preset mutual information entropy screening algorithm to obtain screening fault information;
and carrying out data reconstruction on the initial predicted fault information based on the screening fault information to obtain target predicted fault information, and transmitting the target predicted fault information to a preset data processing terminal to carry out remote monitoring on the target electric appliance.
With reference to the first aspect, in a first implementation manner of the first aspect of the present invention, the performing time sequence feature extraction on the real-time parameter set to obtain a target time sequence feature set includes:
identifying missing data points of the real-time parameter set, and acquiring a plurality of missing data points;
filling missing values of a plurality of missing data points through a preset linear interpolation algorithm to obtain a filling parameter set;
performing outlier replacement on the filling parameter set to obtain a replacement parameter set;
carrying out data smoothing processing on the replacement parameter set through a preset sliding window algorithm to obtain a denoising parameter set;
carrying out correlation analysis on the denoising parameter set to obtain parameter correlation data;
and carrying out time sequence feature extraction on the denoising parameter set based on the parameter correlation data to obtain a target time sequence feature set.
With reference to the first aspect, in a second implementation manner of the first aspect of the present invention, the obtaining a historical parameter set of the target electrical appliance, performing data division on the historical parameter set to obtain a plurality of division data sets, and performing model training on a preset initial SF-ELM model through the plurality of division data sets to obtain a target SF-ELM model, where the method includes:
Acquiring the historical parameter set, and performing data volume traversal on the historical parameter set to obtain a target data volume of the historical parameter set;
performing division range calibration on the historical parameter set through the target data volume to obtain a corresponding data division range;
performing data division on the historical parameter set through the data division range to obtain a plurality of division data sets;
performing model decomposition on the initial SF-ELM model to obtain a plurality of sub-models, and performing model weight analysis on the plurality of sub-models through a plurality of divided data sets to obtain model weight data corresponding to each sub-model;
and carrying out model integration on a plurality of sub-models through model weight data corresponding to each sub-model to obtain the target SF-ELM model.
With reference to the first aspect, in a third implementation manner of the first aspect of the present invention, the inputting the target timing feature set into the target SF-ELM model to perform fault information prediction, to obtain initial predicted fault information includes:
inputting the target time sequence feature set into an input layer of the target SF-ELM model to perform data weight calculation to obtain a corresponding weight set;
Inputting the weight set into a first hidden layer of the target SF-ELM model to construct a parameter matrix, so as to obtain an initial parameter matrix;
inputting the initial parameter matrix into a second hidden layer of the target SF-ELM model to update the matrix to obtain a target electrical appliance matrix;
and inputting the target electrical appliance matrix into an output layer of the target SF-ELM model to perform data residual calculation to obtain target data residual, and generating the initial prediction fault information based on the target data residual.
With reference to the first aspect, in a fourth implementation manner of the first aspect of the present invention, performing spatial correlation analysis on the plurality of sensors based on the initial predicted fault information to obtain spatial correlation data, and performing topology construction on the plurality of sensors based on the spatial correlation data through the initial predicted fault information to obtain a target topology, where the method includes:
carrying out data standardization processing on the initial prediction fault information to obtain standardized data information;
sensor pairing is carried out on the standardized data information, and target standardized data corresponding to each sensor is obtained;
based on target standardized data corresponding to each sensor, performing covariance matrix calculation on each two sensors to obtain a plurality of covariance matrices;
Carrying out space correlation analysis on a plurality of sensors through a plurality of covariance matrixes to obtain space correlation data;
based on a preset correlation threshold, extracting the significant correlation data of the spatial correlation data to obtain significant correlation data;
performing data fusion on the significant related data and the initial prediction fault data to obtain fusion related data;
and constructing topological relations of the plurality of sensors through the fusion related data to obtain the target topological relation.
With reference to the first aspect, in a fifth implementation manner of the first aspect of the present invention, according to the target topological relation, the data filtering, by a preset mutual information entropy filtering algorithm, is performed on the initial predicted fault information to obtain filtered fault information, where the filtering includes:
extracting sensor nodes from the target topological relation to obtain information of a plurality of sensor nodes;
carrying out information association calculation on each sensor node information through a preset mutual information entropy screening algorithm to obtain information association data;
carrying out joint probability density distribution analysis on the information association degree data to obtain a probability density distribution diagram;
Performing curve development trend analysis on the probability density distribution map to obtain a corresponding curve development trend;
calibrating the data screening range of the probability density distribution map through the curve development trend to obtain at least one data screening range;
and carrying out data or o-gram selection on the initial prediction fault information through at least one data screening range to obtain screening fault information.
With reference to the first aspect, in a sixth implementation manner of the first aspect of the present invention, the performing data reconstruction on the initial predicted fault information based on the screening fault information to obtain target predicted fault information, and transmitting the target predicted fault information to a preset data processing terminal to remotely monitor the target electrical appliance, includes:
performing data classification on the initial prediction fault information based on the screening fault information to obtain dominant variable data and auxiliary variable data;
constructing a data mapping relation between the dominant variable data and the auxiliary variable data to obtain a target data mapping relation;
performing data reconstruction on the initial prediction fault information through the target data mapping relation to obtain target prediction fault information;
And transmitting the target prediction fault information to a preset data processing terminal to remotely monitor the target electric appliance.
A second aspect of the present invention provides a remote monitoring system for an electric appliance, the remote monitoring system for an electric appliance comprising:
the acquisition module is used for acquiring real-time parameters of the electric appliance through a plurality of sensors installed in the target electric appliance to obtain a real-time parameter set;
the extraction module is used for extracting time sequence characteristics of the real-time parameter set to obtain a target time sequence characteristic set;
the dividing module is used for acquiring a historical parameter set of the target electrical appliance, carrying out data division on the historical parameter set to obtain a plurality of divided data sets, and carrying out model training on a preset initial SF-ELM model through the plurality of divided data sets to obtain a target SF-ELM model;
the prediction module is used for inputting the target time sequence feature set into the target SF-ELM model to perform fault information prediction to obtain initial prediction fault information;
the construction module is used for carrying out space correlation analysis on the plurality of sensors based on the initial prediction fault information to obtain space correlation data, and carrying out topology relation construction on the plurality of sensors based on the space correlation data to obtain a target topology relation;
The screening module is used for carrying out data screening on the initial prediction fault information through a preset mutual information entropy screening algorithm according to the target topological relation to obtain screening fault information;
and the reconstruction module is used for carrying out data reconstruction on the initial predicted fault information based on the screening fault information to obtain target predicted fault information, and transmitting the target predicted fault information to a preset data processing terminal to carry out remote monitoring on the target electric appliance.
A third aspect of the present application provides a remote monitoring device for an appliance, comprising: a memory and at least one processor, the memory having instructions stored therein; the at least one processor invokes the instructions in the memory to cause the remote monitoring device for an appliance to perform the remote monitoring method for an appliance described above.
A fourth aspect of the present application provides a computer readable storage medium having instructions stored therein which, when run on a computer, cause the computer to perform the above-described remote monitoring method for an appliance.
In the technical scheme provided by the application, the real-time parameters of the electric appliance are acquired through a plurality of sensors installed in the target electric appliance, so that a real-time parameter set is obtained; extracting time sequence characteristics of the real-time parameter set to obtain a target time sequence characteristic set; acquiring a historical parameter set of a target electrical appliance, performing data division on the historical parameter set to obtain a plurality of division data sets, and performing model training on a preset initial SF-ELM model through the plurality of division data sets to obtain a target SF-ELM model; inputting the target time sequence feature set into a target SF-ELM model to predict fault information, and obtaining initial predicted fault information; performing spatial correlation analysis on the plurality of sensors based on the initial prediction fault information to obtain spatial correlation data, and constructing topological relations of the plurality of sensors based on the spatial correlation data to obtain target topological relations; according to the target topological relation, carrying out data screening on the initial prediction fault information through a preset mutual information entropy screening algorithm to obtain screening fault information; and carrying out data reconstruction on the initial predicted fault information based on the screening fault information to obtain target predicted fault information, and transmitting the target predicted fault information to a preset data processing terminal to carry out remote monitoring on the target electric appliance. In the scheme of the application, the state of the electric appliance can be monitored in real time by collecting the real-time parameters and extracting the time sequence characteristics of the target electric appliance through the plurality of sensors. Based on the fault information prediction of the target SF-ELM model, potential electrical appliance faults can be identified in advance, timely maintenance and maintenance measures can be facilitated, and the influence of the faults on the normal operation of the equipment is reduced. Through spatial correlation analysis, the correlation among the sensors can be deeply known, and the data correlation among the sensors is the strongest can be identified. This helps to determine the physical relationships and layout inside the appliance, provides clues about the cause of the failure, and further improves maintenance strategies. Constructing the target topological relation enables the connection mode among various sensors inside the electrical equipment to be better understood. This is important for visualizing and optimizing the structure and performance of the device, helping to improve the efficiency and maintainability of the electrical device. The mutual information entropy screening algorithm is used for screening the predicted fault information, so that information related to the faults of the electrical appliances can be extracted, the redundancy of the information is reduced, and the accuracy of fault detection is improved. The data reconstruction further perfects the fault information, so that the fault information has more information quantity. And the target prediction fault information is transmitted to the data processing terminal, so that remote monitoring and management of the electrical equipment are realized. This means that the equipment operator or engineer can access the equipment status and fault information at any time and any place, and make decisions quickly, improving the reliability and safety of the equipment. So as to further improve the accuracy of remote monitoring of the electric appliance.
Drawings
FIG. 1 is a schematic diagram of an embodiment of a remote monitoring method for an electrical appliance according to an embodiment of the present invention;
FIG. 2 is a flow chart of data partitioning of a historical parameter set according to an embodiment of the present invention;
FIG. 3 is a flowchart of inputting a target timing sequence feature set into a target SF-ELM model for fault information prediction in an embodiment of the present invention;
FIG. 4 is a flow chart of spatial correlation analysis for multiple sensors in an embodiment of the invention;
FIG. 5 is a schematic diagram of one embodiment of a remote monitoring system for an appliance in an embodiment of the present invention;
fig. 6 is a schematic diagram of an embodiment of a remote monitoring apparatus for an electric appliance according to an embodiment of the present invention.
Detailed Description
The embodiment of the invention provides a remote monitoring method and a remote monitoring system for an electric appliance, which are used for improving the accuracy of remote monitoring of the electric appliance.
The terms "first," "second," "third," "fourth" and the like in the description and in the claims and in the above drawings, if any, are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments described herein may be implemented in other sequences than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed or inherent to such process, method, article, or apparatus.
For ease of understanding, a specific flow of an embodiment of the present invention is described below with reference to fig. 1, and an embodiment of a remote monitoring method for an electrical appliance in an embodiment of the present invention includes:
s101, acquiring real-time parameters of an electric appliance through a plurality of sensors installed in a target electric appliance to obtain a real-time parameter set;
it is to be understood that the execution subject of the present invention may be a remote monitoring system for an electric appliance, and may also be a terminal or a server, which is not limited herein. The embodiment of the invention is described by taking a server as an execution main body as an example.
Specifically, the server selects an appropriate type and number of sensors to meet the monitoring needs of the target appliance. These sensors may include temperature sensors, humidity sensors, pressure sensors, current sensors, light sensitive sensors, etc., with the specific choice depending on the nature and range of parameters desired. The sensors need to be mounted precisely at critical locations on the electrical equipment to ensure accurate parameter acquisition. These locations are typically where the most representative data can be provided to reflect the actual state of the device. A data acquisition system is established to connect the sensor to a data acquisition device (such as a data acquisition card or sensor node) and to configure the data acquisition software. The sensor will constantly collect data and transmit it to the data collection device, which is then responsible for receiving, processing and storing the data. To ensure reliability and integrity of data, redundancy and fault recovery mechanisms are typically introduced in the system, such as using multiple sensors to measure the same parameters to detect sensor faults or data anomalies. The collected data can be used for monitoring the performance of the target electrical appliance in real time, and helping a user or an operator to identify potential problems and take corresponding measures through remote access and data analysis so as to maintain the reliability and the safety of the equipment.
S102, extracting time sequence features of the real-time parameter set to obtain a target time sequence feature set;
specifically, the server processes missing data points in the real-time parameter set. These missing data points are lost due to sensor failure, communication problems, or other reasons. To address this problem, the server uses data recognition techniques to detect the presence of missing data points and populates the data points by a linear interpolation algorithm. Linear interpolation uses the linear relationship between known data points to estimate the value of the missing points to make the data set continuous. For example, if the server has a time series data set in which data at certain points in time is missing, linear interpolation will evaluate the value of the missing point by knowing the value of the point and time, thus populating the data set. Abnormal values are detected and replaced. Outliers interfere with the accuracy and reliability of the data and therefore need to be processed. Common outlier detection methods include statistical methods, threshold-based methods, and machine learning methods. When an outlier is detected, the data set may be repaired by replacing it with the appropriate value. For example, if a sensor reading is abnormally high, it may be replaced with an average or median value of the sensor when it is operating normally. Data often contains noise and fluctuations that interfere with analysis of the temporal characteristics. Therefore, the server performs data smoothing processing using a sliding window algorithm. The sliding window algorithm divides the data set into a plurality of time windows and then calculates an average or other statistical feature within each window. This helps to reduce high frequency noise in the data, making the timing characteristics more stable. For example, for temperature data, the server calculates the average temperature per hour, rather than using raw data per minute. The server then performs a parameter correlation analysis. This step aims at identifying the correlation between the parameters in order to better understand the behaviour of the appliance. The correlation analysis may employ statistical methods such as pearson correlation coefficients or machine learning based methods. For example, if the server monitors a motor, the correlation between current and vibration may be analyzed to determine if there is a pattern or correlation between them. Based on the parameter correlation data, the server extracts a set of target timing characteristics. These features may include statistical features (e.g., mean, standard deviation), frequency domain features (e.g., frequency components), time domain features (e.g., peak, waveform shape), and other high-level features. Extracting timing characteristics facilitates deeper analysis and prediction of the performance and status of the appliance. For example, the server collects temperature, current and vibration data for the device. In the time sequence feature extraction process, the server finds that some data points are missing, some data have abnormal values, and the fluctuation of the data is large. The server fills in missing data points using a linear interpolation algorithm, making the data set continuous. The server detects and replaces outliers to ensure accuracy of the data. The server performs sliding window smoothing on the data to reduce high frequency noise. The server performs parameter correlation analysis to find that there is a certain correlation between the temperature and the current, which indicates that the current will rise correspondingly when the temperature rises. Based on these analysis results, the server extracts a series of timing characteristics such as average temperature, current peak, temperature-current correlation, and so on. These features can be used for further fault prediction, performance optimization or maintenance decisions to ensure reliable and stable operation of the industrial equipment.
S103, acquiring a historical parameter set of a target electrical appliance, performing data division on the historical parameter set to obtain a plurality of division data sets, and performing model training on a preset initial SF-ELM model through the plurality of division data sets to obtain a target SF-ELM model;
specifically, the server obtains a history parameter set of the target appliance. These historical parameters typically include various parameter data of the appliance over a period of time, such as temperature, humidity, current, etc. The data can be collected and stored in real time by the sensor or can be retrieved by the history record. The server performs data partitioning on the set of historical parameters for model training. The server determines a target amount of data, i.e., the amount of historical data that the server is to use for training. This may be selected according to requirements, typically determined according to the availability of data and training requirements of the model. And (5) demarcating the data division range. This involves selecting which time periods or data windows in the historical parameter set to use for training. For example, if the server wants to train a model that predicts future performance of the appliance, the server will select the data for the last period of time. The server divides the set of history parameters into a plurality of data subsets according to the data division range. Each subset contains a different time period or window of historical parameter data. These subsets will be used to train different parts of the model so that the model can capture data patterns and changes for different periods of time. The server performs model decomposition on the initial SF-ELM model, and splits the initial SF-ELM model into a plurality of sub-models. These sub-models may be the same type of model or may be different types of models depending on the complexity of the problem and the requirements of the model. Each sub-model will handle the training task of one subset of the data partitions exclusively. After obtaining the sub-models, the server performs a model weight analysis on each sub-model through a plurality of data partitioning subsets. The server evaluates the performance and behavior of each sub-model on different subsets of data. This may be achieved by an index such as a mean square error, accuracy or other correlation index. The model weight data represents the importance of each sub-model on different data partitions. And the server performs model integration through model weight data corresponding to each sub-model to obtain a target SF-ELM model. Model integration may employ different methods such as weighted averaging, voting, or stacking. The goal of the integrated model is to combine the predictions of multiple sub-models to improve overall predictive performance and generalization capability. For example, the device includes a temperature sensor, a humidity sensor, and a current sensor, and the server trains an SF-ELM model to predict future performance of the device. The server has one year of historical parameter data available. The server decides to train the model using the last three months of data, which is the data partitioning range of the server. The server divides the data during this period into subsets, each subset representing one month of data. The server then breaks down the initial SF-ELM model into three sub-models, each of which will be trained on one month of data, respectively. The server obtains model weight data of each sub-model by evaluating the performance of each sub-model on different month data. For example, a sub-model performs better in the first month, so its weight will be higher in the integrated model. The server uses these model weight data to integrate the three sub-models, resulting in the target SF-ELM model. The integrated model predicts the device performance by using the prediction results of all sub-models, thereby improving the accuracy and reliability of the model. This model can be used for remote monitoring and real-time decision making to ensure proper operation of the industrial equipment.
S104, inputting the target time sequence feature set into a target SF-ELM model to predict fault information, and obtaining initial predicted fault information;
specifically, the server inputs the set of target timing characteristics to an input layer of the target SF-ELM model. In this step, the model will calculate the weight of each feature to determine its degree of influence on the model output. These weights represent the importance of each feature and may be calculated by various methods such as random initialization, recursive Least Squares (RLS), gradient descent, etc. The weight set will be used for the next parameter matrix construction. The server inputs the set of weights into a first hidden layer of the target SF-ELM model to construct an initial parameter matrix. This parameter matrix contains information of the feature weights and is passed as input to the second hidden layer. The construction of a parameter matrix can be seen as a linear combination of features, where the weight of each feature is multiplied by the corresponding time series feature and the results are summarized into the parameter matrix. The initial parameter matrix is then input to the second hidden layer of the target SF-ELM model for matrix updating. At this level, the model may apply different activation functions, nonlinear mapping, or other techniques to further extract and process feature information. The purpose of this step is to make the data more suitable for fault information prediction by nonlinear transformation. And the target electrical appliance matrix, namely the parameter matrix after matrix updating, is input to an output layer of the target SF-ELM model to perform data residual calculation. At the output layer, the model will calculate a predicted value from the input target appliance matrix and then compare it with the actual observed value to calculate the data residual. The data residual is the difference between the predicted and actual values of the model, which represents the prediction error or residual of the model. Based on the calculated data residuals, the server generates initially predicted fault information. The data residuals typically reflect the deviation between the state of the target appliance and the model predictions, which can be used to identify if an appliance is abnormal or potentially faulty. For example, if the data residuals increase significantly, indicating a problem with the performance of the appliance, further inspection and maintenance is required.
S105, carrying out spatial correlation analysis on the plurality of sensors based on the initial prediction fault information to obtain spatial correlation data, and constructing topological relations of the plurality of sensors based on the spatial correlation data to obtain target topological relations;
specifically, the server performs data normalization processing on the initially predicted failure information. Normalization can allow the data between different sensors to have the same dimensions and ranges to ensure that they can be effectively compared and analyzed. The server pairs between each sensor. The server combines the sensors two by two to study the relationship between them. For example, if the server has 4 sensors, 6 different sensor pairs may be formed. For each sensor pairing, the server calculates a covariance matrix between them. The covariance matrix is used to measure the linear relationship and correlation between two sensors. By calculating the covariance matrix, the server knows whether the data changes between the sensors have correlation. Based on the plurality of covariance matrices, the server performs a spatial correlation analysis to evaluate the degree of correlation between the sensors. This step helps to determine which sensors have significant spatial correlation between them, indicating that they play an important role in fault detection and prediction. The server presets a correlation threshold for determining which correlations are significant. The choice of the correlation threshold depends on the specific application scenario and the problem to be solved. In general, a significant correlation can be considered if the absolute value of the correlation is greater than or equal to a threshold. The server extracts significant correlation data from the spatial correlation data according to the correlation threshold. These significantly related data represent a strong association between sensors, related to fault information of the appliance. The server performs data fusion on the significantly relevant data and the initially predicted fault information. The data fusion can adopt various methods, such as weighted average or logic operation, to combine the information of different data sources together so as to improve the accuracy and reliability of fault information. Based on the fusion related data, the server builds a topological relationship between the sensors. The topological relation represents the connection and influence relation among the sensors, and is helpful for the server to understand the structure and operation mode of the electrical system. This topology can be used for further fault detection, prediction and maintenance decisions. For example, assume that a plurality of sensors on one production line, including a temperature sensor, a humidity sensor, a pressure sensor, and the like. The server predicts the fault condition of the production line by analyzing the spatial correlation between the sensors. The server pairs each sensor, resulting in a plurality of sensor pairs. The server calculates the covariance matrix between each pair of sensors to learn the linear relationship between them. The server then performs a spatial correlation analysis, which finds that there is a significant correlation between the temperature sensor and the humidity sensor, meaning that changes in humidity can affect the temperature. The server sets a correlation threshold and extracts significantly correlated data. The server performs data fusion on the significantly related data and the initial predicted fault information to obtain more accurate fault information. Based on these data, the server builds a topological graph between the sensors, helping the server understand the structure and failure modes of the production line.
S106, data screening is carried out on the initial prediction fault information through a preset mutual information entropy screening algorithm according to the target topological relation, and screening fault information is obtained;
specifically, the server extracts sensor node information from the target topology. These nodes represent locations or components associated with sensors in the appliance system. Each node may have one or more sensors associated with it. And the server uses a preset mutual information entropy screening algorithm to calculate the information association degree of each sensor node information. The information relevance measures the relevance between each sensor node and the fault information. This may be achieved by calculating mutual information entropy or other correlation indicators. Based on the information association data, the server performs probability density distribution analysis. This step is intended to understand the distribution of each sensor node information to determine which node information has a higher information density and predictability. The probability density distribution may be generally represented by a histogram or a kernel density estimation. And on the basis of probability density distribution, the server performs curve development trend analysis. The server observes historical trends of changes in each of the sensor node information to determine if a pattern or trend associated with the fault exists. This can be achieved by plotting a time series graph and performing trend analysis. Based on the curve development trend and the probability density distribution, the server calibrates at least one data screening range. This range defines which fault information data points should be retained and which should be culled. The data screening range typically includes upper and lower limits representing a range of fault information values. The server compares the initially predicted fault information with at least one data screening range for data screening. Only the fault information data points within the range are retained and the remaining data points are deleted or marked as untrusted. In this way, the server obtains filtered fault information, which includes information highly related to the sensor nodes. For example, a plurality of sensors including a temperature sensor, a humidity sensor, and a vibration sensor. The server uses these sensors to predict faults on the production line. The server extracts sensor node information from the topology of the production line, including the location and type of each sensor. Then, the server calculates the information association degree of each sensor node information and fault information by using a mutual information entropy screening algorithm. The server analyzes the probability density distribution and knows the distribution condition of the information of each sensor node. In the trend analysis of the curve, the server plots the time series of temperature, humidity and vibration sensors and checks whether an abnormal pattern or trend exists. Based on the results of the analysis, the server calibrates the data screening range, e.g., for a temperature sensor, the server sets a temperature range to hold data points highly correlated to fault information. And the server compares the initial predicted fault information with a data screening range, and reserves data points in the range to obtain the screened fault information. Such information will more accurately reflect fault information highly correlated to the sensor nodes, facilitating remote monitoring and prediction of production line faults.
And S107, carrying out data reconstruction on the initial predicted fault information based on the screening fault information to obtain target predicted fault information, and transmitting the target predicted fault information to a preset data processing terminal to carry out remote monitoring on the target electric appliance.
The server classifies the initial predicted failure information based on the screening failure information. The servers divide the fault information into two categories: dominant variable data and auxiliary variable data. The primary variable is typically core data directly related to the appliance fault, while the secondary variable is data related to appliance performance monitoring but not directly related to the fault. And the server builds a data mapping relation between the main variable data and the auxiliary variable data. The purpose of this step is to determine the relationship between the dominant and auxiliary variables in order to better understand the interactions between them. The data mapping relationship may be linear or non-linear and may be established using various statistical and machine learning techniques. Based on the data mapping relation, the server performs data reconstruction on the initial prediction fault information to obtain target prediction fault information. The server combines and transforms the primary and secondary variable data to obtain more accurate and comprehensive fault information. The data reconstruction can be performed by adopting linear regression, principal component analysis, neural network and other technologies. And the server transmits the target prediction fault information to a preset data processing terminal so as to carry out remote monitoring. The data processing terminal may be a cloud server, a monitoring center, or other data processing device. Here, the monitoring personnel can receive and analyze the target predictive failure information and take corresponding measures, such as maintenance of the appliance or taking precautions, as needed to ensure proper operation of the appliance. Consider, for example, a motor. Servers have collected data regarding motor performance and health, including current, vibration, temperature, etc., using various sensors. The server wishes to use this data to predict motor failure and to remotely monitor. The server analyzes the collected data and classifies the data into dominant variable data and auxiliary variable data. Current and vibration data are the dominant variables, while temperature data are the auxiliary variables. The server uses a data map construction method to determine the relationship between current and vibration, and their relationship to motor faults. This may be achieved by modeling or using correlation analysis. Based on the data mapping relation, the server reconstructs fault information, and combines the dominant variable data with the auxiliary variable data to obtain more comprehensive motor health state information. And the server transmits the reconstructed information to a cloud server, a monitoring center or a data processing terminal. Here, the monitoring personnel can receive the health status information of the motor in real time, and monitor whether potential fault risks exist. If the system detects an anomaly, the monitoring personnel may take appropriate action, such as raising an alarm, scheduling maintenance or adjusting operating parameters to ensure reliable operation of the motor.
In the embodiment of the invention, the state of the electric appliance can be monitored in real time by collecting the real-time parameters and extracting the time sequence characteristics of the target electric appliance through the plurality of sensors. Based on the fault information prediction of the target SF-ELM model, potential electrical appliance faults can be identified in advance, timely maintenance and maintenance measures can be facilitated, and the influence of the faults on the normal operation of the equipment is reduced. Through spatial correlation analysis, the correlation among the sensors can be deeply known, and the data correlation among the sensors is the strongest can be identified. This helps to determine the physical relationships and layout inside the appliance, provides clues about the cause of the failure, and further improves maintenance strategies. Constructing the target topological relation enables the connection mode among various sensors inside the electrical equipment to be better understood. This is important for visualizing and optimizing the structure and performance of the device, helping to improve the efficiency and maintainability of the electrical device. The mutual information entropy screening algorithm is used for screening the predicted fault information, so that information related to the faults of the electrical appliances can be extracted, the redundancy of the information is reduced, and the accuracy of fault detection is improved. The data reconstruction further perfects the fault information, so that the fault information has more information quantity. And the target prediction fault information is transmitted to the data processing terminal, so that remote monitoring and management of the electrical equipment are realized. This means that the equipment operator or engineer can access the equipment status and fault information at any time and any place, and make decisions quickly, improving the reliability and safety of the equipment. So as to further improve the accuracy of remote monitoring of the electric appliance.
In a specific embodiment, the process of executing step S102 may specifically include the following steps:
(1) Identifying missing data points of the real-time parameter set, and acquiring a plurality of missing data points;
(2) Filling missing values of a plurality of missing data points through a preset linear interpolation algorithm to obtain a filling parameter set;
(3) Performing outlier replacement on the filling parameter set to obtain a replacement parameter set;
(4) Carrying out data smoothing processing on the replacement parameter set through a preset sliding window algorithm to obtain a denoising parameter set;
(5) Carrying out correlation analysis on the denoising parameter set to obtain parameter correlation data;
(6) And carrying out time sequence feature extraction on the denoising parameter set based on the parameter correlation data to obtain a target time sequence feature set.
Specifically, the server identifies missing data points for the real-time parameter set. Missing data points refer to data points that have not been collected for various reasons, such as sensor failure or communication problems. Identifying missing data points may be accomplished by observing null values in the data set or using a missing value detection algorithm. The server fills in these missing data points using a preset linear interpolation algorithm. The linear interpolation algorithm may estimate the value of the missing data point from the known data points. For example, if the data point between times t1 and t2 is known, the server uses linear interpolation to estimate the data value between times t1 and t 2. After filling, the server performs outlier replacement on the filling parameter set. Outliers refer to data points that are significantly unreasonable or extreme compared to other data points. Outliers can interfere with subsequent data analysis and modeling. The server uses anomaly detection algorithms to identify and replace these anomalies to ensure the quality and consistency of the data. And the server performs data smoothing processing on the replacement parameter set through a preset sliding window algorithm. Data smoothing helps to remove noise and fluctuations in the data, making the data more readable and stable. The sliding window algorithm will slide a window over the data sequence and use the data points within the window to calculate an average or other smoothed statistic. Based on the smoothed data, the server performs a parameter correlation analysis. Parameter correlation analysis helps to understand the relationship between different parameters, including linear and non-linear relationships. This may be achieved by calculating correlation coefficients, covariance matrices, or using machine learning methods. Based on the parameter correlation data, the server performs timing characteristic extraction on the denoising parameter set. The timing characteristics are information extracted from time series data and can be used for modeling and fault prediction. These features may include mean, standard deviation, trend, periodicity, etc.
In a specific embodiment, as shown in fig. 2, the process of performing step S103 may specifically include the following steps:
s201, acquiring a historical parameter set, and performing data volume traversal on the historical parameter set to obtain a target data volume of the historical parameter set;
s202, calibrating a dividing range of a historical parameter set through a target data volume to obtain a corresponding data dividing range;
s203, carrying out data division on the historical parameter set through a data division range to obtain a plurality of division data sets;
s204, performing model decomposition on the initial SF-ELM model to obtain a plurality of sub-models, and performing model weight analysis on the plurality of sub-models through a plurality of divided data sets to obtain model weight data corresponding to each sub-model;
s205, performing model integration on the plurality of sub-models through model weight data corresponding to each sub-model to obtain a target SF-ELM model.
It should be noted that, a set of historical parameters is obtained, and these parameters may be the electrical performance data collected previously. And determining the target data volume according to the requirements of the application. The target data amount represents the number of data points that the server uses for model training. And according to the target data volume, the server calculates the data dividing range of the history parameter set. This range may represent which data points in the historical data set will be used for model training and which will be used for validating or testing the model. Typically, the server divides the data into training, validation and test sets to evaluate model performance. Using the data partitioning range, the server partitions the historical parameter set into a plurality of partitioned data sets. These data sets include data for training, data for validation, and data for testing. Such partitioning helps to evaluate the performance of the model and avoid overfitting. The server performs model decomposition on the initial SF-ELM model. SF-ELM models typically contain multiple sub-models, each for processing different aspects of data or tasks. Model decomposition may divide the model into multiple sub-portions as needed, each with its own inputs and outputs. The server performs model weight analysis on the plurality of sub-models by a plurality of partitioned data sets. Model weights are parameters used to adjust the importance of each sub-model. These weights may be assigned according to the performance of the different data sets to ensure that each sub-model contributes to the overall performance of the model. And carrying out model integration on the plurality of sub-models by the server through model weight data corresponding to each sub-model to obtain the target SF-ELM model. The model integration can adopt methods such as weighted average, voting method, stacking and the like to obtain a more powerful and robust model, and can be better suitable for different data and tasks.
In a specific embodiment, as shown in fig. 3, the process of executing step S104 may specifically include the following steps:
s301, inputting a target time sequence feature set into an input layer of a target SF-ELM model to perform data weight calculation to obtain a corresponding weight set;
s302, inputting a weight set into a first hidden layer of a target SF-ELM model to construct a parameter matrix, and obtaining an initial parameter matrix;
s303, inputting the initial parameter matrix into a second hidden layer of the target SF-ELM model to update the matrix to obtain a target electrical appliance matrix;
s304, inputting the target appliance matrix into an output layer of the target SF-ELM model to perform data residual calculation to obtain target data residual, and generating initial prediction fault information based on the target data residual.
The server inputs the target time sequence feature set into the input layer of the SF-ELM model. At the input level, each feature is computed with a corresponding weight. The weights can be seen as the degree of contribution of each feature to the model, which are used to weight the feature values in order to better capture the pattern and relationship of the data. And inputting the calculated weight set into a first hidden layer of the SF-ELM model. At the first hidden layer, the weights are multiplied by the input features and form a parameter matrix. This parameter matrix contains the initial parameters of the model, which will be optimized to fit the data during model training. The parameter matrix is further passed to a second hidden layer of the SF-ELM model. At the second hidden layer, the parameter matrix will be further updated, typically by regularization or other optimization algorithms, to improve the performance of the model. The updated parameter matrix may better capture the complexity and pattern of the data. The updated parameter matrix is transferred to the output layer of the SF-ELM model. At the output layer, the model uses the updated parameter matrix to linearly combine the input data and generate the predicted data. The difference from the actual data is called the data residual. The data residuals represent how well the model fits, which are a measure of the model's performance. Based on the data residuals, initial prediction fault information may be generated. The larger the data residual, the better the model is not fit to the data, and there are potential faults or anomalies. Depending on the size and direction of the data residuals, problems with the appliance, such as performance degradation or failure, may be predicted.
In a specific embodiment, as shown in fig. 4, the process of performing step S105 may specifically include the following steps:
s401, carrying out data standardization processing on initial prediction fault information to obtain standardized data information;
s402, carrying out sensor pairing on the standardized data information to obtain target standardized data corresponding to each sensor;
s403, based on target standardized data corresponding to each sensor, performing covariance matrix calculation on each two sensors to obtain a plurality of covariance matrices;
s404, performing spatial correlation analysis on a plurality of sensors through a plurality of covariance matrixes to obtain spatial correlation data;
s405, extracting significant relevant data from the spatial relevant data based on a preset relevance threshold value to obtain significant relevant data;
s406, carrying out data fusion on the significant related data and the initial prediction fault data to obtain fusion related data;
s407, constructing topological relations of the plurality of sensors by fusing the related data to obtain a target topological relation.
Specifically, the data normalization processing is performed on the initial prediction failure information. Data normalization is the conversion of data into a standard normal distribution with a mean of 0 and standard deviation of 1 to ensure that the data has similar dimensions and ranges of variation. This can be achieved by subtracting the mean and dividing by the standard deviation. And carrying out sensor pairing on the standardized data information. Sensor pairing can determine which sensors should be paired based on similarity of data features or physical relationships. For example, if two sensors measure similar physical quantities, such as temperature and humidity, they may be paired. For each paired pair of sensors, a covariance matrix is calculated between them. The covariance matrix reflects the degree of linear relationship between the two sensors. By calculating the covariance matrix, the server knows whether there is a correlation between the sensors and the strength and direction of the correlation. A spatial correlation analysis is performed on the plurality of sensors by a plurality of covariance matrices. The spatial correlation analysis may be implemented by computing eigenvalues and eigenvectors of the covariance matrix. The eigenvalue represents the intensity of the correlation and the eigenvector represents the direction of the correlation. The spatial correlation analysis helps determine which sensors have spatial correlation between them and provides information for subsequent topological construction. And performing significant correlation data extraction on the spatial correlation data based on a preset correlation threshold. Only correlations with sufficient strength will be considered significant correlations. This can be achieved by setting a correlation threshold value, only pairs of sensors whose correlation exceeds the threshold value are retained. And carrying out data fusion on the significantly relevant data and the initial predicted fault data. The data fusion may employ weighted averaging, logical operations, or other methods to combine the correlation information with the fault information to generate fusion-related data. And constructing topological relations of the plurality of sensors by fusing the related data. The topological relation represents the connection mode and network structure between the sensors. It may be determined by analyzing the correlation between the sensors, for example, if two sensors are highly correlated, they have a connection relationship.
In a specific embodiment, the process of executing step S106 may specifically include the following steps:
(1) Extracting sensor nodes from the target topological relation to obtain information of a plurality of sensor nodes;
(2) Carrying out information association calculation on each sensor node information through a preset mutual information entropy screening algorithm to obtain information association data;
(3) Carrying out joint probability density distribution analysis on the information association degree data to obtain a probability density distribution diagram;
(4) Carrying out curve development trend analysis on the probability density distribution diagram to obtain a corresponding curve development trend;
(5) Calibrating a data screening range of the probability density distribution map through a curve development trend to obtain at least one data screening range;
(6) And carrying out data or treatment on the initial prediction fault information through at least one data screening range to obtain screening fault information.
Specifically, a plurality of sensor node information is extracted from the target topology. The sensor nodes are independent measuring points in the sensor network and can be different sensors on industrial equipment or sensors on household appliances. These nodes are the basic building blocks of the network for monitoring and collecting data. And calculating the information association degree of each sensor node information by using a preset mutual information entropy screening algorithm. The information correlation degree indicates a degree of information correlation between the sensor nodes. Mutual information entropy is a method for measuring the correlation between random variables that can help determine which sensor nodes have a high degree of correlation between them. And carrying out joint probability density distribution analysis on the information association degree data. This step aims at knowing the distribution of the information association, i.e. which sensor nodes have higher information association and which have lower information association. The probability density profile may help visualize these correlations. And (5) carrying out curve development trend analysis on the probability density distribution map. By observing the curve shape and the change trend in the probability density distribution diagram, the change trend of the information association degree between the sensor nodes with time or other factors can be identified. These trends facilitate further data screening. At least one data screening range is determined based on the trend of the curve. The data filtering range refers to a range of information association values, and is used for filtering fault information. The range can be calibrated according to the development trend of the curve and specific application requirements. The initial predicted fault information is filtered using at least one data filtering range. Only information association values within a specified range are considered significantly relevant and thus remain as filtered fault information.
In a specific embodiment, the process of executing step S107 may specifically include the following steps:
(1) Data classification is carried out on the initial prediction fault information based on the screening fault information, and leading variable data and auxiliary variable data are obtained;
(2) Constructing a data mapping relation between the main variable data and the auxiliary variable data to obtain a target data mapping relation;
(3) Carrying out data reconstruction on the initial prediction fault information through a target data mapping relation to obtain target prediction fault information;
(4) And transmitting the target prediction fault information to a preset data processing terminal to remotely monitor the target electric appliance.
Specifically, the initial predicted fault information is data classified based on the screening fault information. Fault information is divided into two categories: dominant variable data and auxiliary variable data. The dominant variable data typically contains the most important and most directly fault-related information, while the auxiliary variable data contains auxiliary information, information that provides context or auxiliary analysis in fault diagnosis. And constructing a data mapping relation between the main variable data and the auxiliary variable data. The relationship between the primary and secondary variable data is determined, and how to map them onto the state or performance parameters of the target appliance. The data mapping relationship may be constructed based on a physical model, empirical rules, or a machine learning algorithm. And carrying out data reconstruction on the initial prediction fault information through the target data mapping relation. The classified fault information is converted into the state or performance parameters of the target electric appliance, so that the target prediction fault information is generated. The data reconstruction may be a mathematical model or algorithm that converts the dominant variable data and the auxiliary variable data into estimates of the target parameters. And transmitting the target prediction fault information to a preset data processing terminal so as to realize remote monitoring of the target electric appliance. The data processing terminal may be a cloud server, monitoring system or other remote device for receiving, storing, analyzing and visualizing the predicted fault information. Remote monitoring allows an operator or engineer to monitor the status of the target appliance in real time and take the necessary actions to prevent or address potential faults.
The foregoing describes a remote monitoring method for an electrical appliance in an embodiment of the present invention, and the following describes a remote monitoring system for an electrical appliance in an embodiment of the present invention, referring to fig. 5, an embodiment of the remote monitoring system for an electrical appliance in an embodiment of the present invention includes:
the acquisition module 501 is used for acquiring real-time parameters of the electric appliance through a plurality of sensors installed in the target electric appliance to obtain a real-time parameter set;
the extracting module 502 is configured to perform time sequence feature extraction on the real-time parameter set to obtain a target time sequence feature set;
a dividing module 503, configured to obtain a historical parameter set of the target electrical appliance, divide the historical parameter set into a plurality of divided data sets, and perform model training on a preset initial SF-ELM model through the plurality of divided data sets to obtain a target SF-ELM model;
the prediction module 504 is configured to input the target timing sequence feature set into the target SF-ELM model to perform fault information prediction, so as to obtain initial predicted fault information;
the construction module 505 is configured to perform spatial correlation analysis on the plurality of sensors based on the initial prediction fault information to obtain spatial correlation data, and perform topology relationship construction on the plurality of sensors based on the spatial correlation data through the initial prediction fault information to obtain a target topology relationship;
The screening module 506 is configured to perform data screening on the initial predicted fault information according to the target topological relation by using a preset mutual information entropy screening algorithm, so as to obtain screening fault information;
and the reconstruction module 507 is configured to perform data reconstruction on the initial predicted fault information based on the screening fault information to obtain target predicted fault information, and transmit the target predicted fault information to a preset data processing terminal to perform remote monitoring on the target electrical appliance.
Through the cooperative cooperation of the components, the real-time parameters of the electric appliance are acquired through a plurality of sensors installed in the target electric appliance, and a real-time parameter set is obtained; extracting time sequence characteristics of the real-time parameter set to obtain a target time sequence characteristic set; acquiring a historical parameter set of a target electrical appliance, performing data division on the historical parameter set to obtain a plurality of division data sets, and performing model training on a preset initial SF-ELM model through the plurality of division data sets to obtain a target SF-ELM model; inputting the target time sequence feature set into a target SF-ELM model to predict fault information, and obtaining initial predicted fault information; performing spatial correlation analysis on the plurality of sensors based on the initial prediction fault information to obtain spatial correlation data, and constructing topological relations of the plurality of sensors based on the spatial correlation data to obtain target topological relations; according to the target topological relation, carrying out data screening on the initial prediction fault information through a preset mutual information entropy screening algorithm to obtain screening fault information; and carrying out data reconstruction on the initial predicted fault information based on the screening fault information to obtain target predicted fault information, and transmitting the target predicted fault information to a preset data processing terminal to carry out remote monitoring on the target electric appliance. In the scheme of the application, the state of the electric appliance can be monitored in real time by collecting the real-time parameters and extracting the time sequence characteristics of the target electric appliance through the plurality of sensors. Based on the fault information prediction of the target SF-ELM model, potential electrical appliance faults can be identified in advance, timely maintenance and maintenance measures can be facilitated, and the influence of the faults on the normal operation of the equipment is reduced. Through spatial correlation analysis, the correlation among the sensors can be deeply known, and the data correlation among the sensors is the strongest can be identified. This helps to determine the physical relationships and layout inside the appliance, provides clues about the cause of the failure, and further improves maintenance strategies. Constructing the target topological relation enables the connection mode among various sensors inside the electrical equipment to be better understood. This is important for visualizing and optimizing the structure and performance of the device, helping to improve the efficiency and maintainability of the electrical device. The mutual information entropy screening algorithm is used for screening the predicted fault information, so that information related to the faults of the electrical appliances can be extracted, the redundancy of the information is reduced, and the accuracy of fault detection is improved. The data reconstruction further perfects the fault information, so that the fault information has more information quantity. And the target prediction fault information is transmitted to the data processing terminal, so that remote monitoring and management of the electrical equipment are realized. This means that the equipment operator or engineer can access the equipment status and fault information at any time and any place, and make decisions quickly, improving the reliability and safety of the equipment. So as to further improve the accuracy of remote monitoring of the electric appliance.
The remote monitoring system for an electrical appliance in the embodiment of the present invention is described in detail above in fig. 5 from the point of view of a modularized functional entity, and the remote monitoring device for an electrical appliance in the embodiment of the present invention is described in detail below from the point of view of hardware processing.
Fig. 6 is a schematic structural diagram of a remote monitoring device for an electrical appliance according to an embodiment of the present invention, where the remote monitoring device 600 for an electrical appliance may have a relatively large difference due to configuration or performance, and may include one or more processors (CPU) 610 (e.g., one or more processors) and a memory 620, and one or more storage media 630 (e.g., one or more mass storage devices) storing applications 633 or data 632. Wherein the memory 620 and the storage medium 630 may be transitory or persistent storage. The program stored on the storage medium 630 may include one or more modules (not shown), each of which may include a series of instruction operations in the remote monitoring apparatus 600 for an electric appliance. Still further, the processor 610 may be configured to communicate with the storage medium 630 to execute a series of instruction operations in the storage medium 630 on the remote monitoring device 600 for an appliance.
The remote monitoring device 600 for an appliance may also include one or more power supplies 640, one or more wired or wireless network interfaces 650, one or more input/output interfaces 660, and/or one or more operating systems 631, such as WindowsServe, macOSX, unix, linux, freeBSD, and the like. It will be appreciated by those skilled in the art that the remote monitoring device structure for an appliance shown in fig. 6 does not constitute a limitation of the remote monitoring device for an appliance, and may include more or fewer components than shown, or may combine certain components, or may be arranged in a different arrangement of components.
The invention also provides a remote monitoring device for an electrical appliance, which comprises a memory and a processor, wherein the memory stores computer readable instructions, and the computer readable instructions, when executed by the processor, cause the processor to execute the steps of the remote monitoring method for an electrical appliance in the above embodiments.
The present invention also provides a computer readable storage medium, which may be a non-volatile computer readable storage medium, and may also be a volatile computer readable storage medium, in which instructions are stored which, when executed on a computer, cause the computer to perform the steps of the remote monitoring method for an electrical appliance.
It will be clear to those skilled in the art that, for convenience and brevity of description, specific working procedures of the above-described systems, apparatuses and units may refer to corresponding procedures in the foregoing method embodiments, which are not repeated herein.
The integrated units, if implemented in the form of software functional units and sold or passed as separate products, may be stored in a computer readable storage medium. Based on the understanding that the technical solution of the present invention may be embodied in essence or in a part contributing to the prior art or in whole or in part in the form of a software product stored in a storage medium, comprising instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, or other various media capable of storing program codes.
The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (10)
1. A remote monitoring method for an appliance, the remote monitoring method for an appliance comprising:
acquiring real-time parameters of the electric appliance through a plurality of sensors installed in the target electric appliance to obtain a real-time parameter set;
extracting time sequence characteristics of the real-time parameter set to obtain a target time sequence characteristic set;
acquiring a historical parameter set of the target electrical appliance, performing data division on the historical parameter set to obtain a plurality of division data sets, and performing model training on a preset initial SF-ELM model through the plurality of division data sets to obtain a target SF-ELM model;
inputting the target time sequence feature set into the target SF-ELM model to predict fault information, and obtaining initial predicted fault information;
performing spatial correlation analysis on a plurality of sensors based on the initial prediction fault information to obtain spatial correlation data, and performing topology relation construction on the plurality of sensors based on the spatial correlation data to obtain a target topology relation;
according to the target topological relation, carrying out data screening on the initial prediction fault information through a preset mutual information entropy screening algorithm to obtain screening fault information;
And carrying out data reconstruction on the initial predicted fault information based on the screening fault information to obtain target predicted fault information, and transmitting the target predicted fault information to a preset data processing terminal to carry out remote monitoring on the target electric appliance.
2. The remote monitoring method for electrical appliances according to claim 1, wherein the performing the time sequence feature extraction on the real-time parameter set to obtain a target time sequence feature set comprises:
identifying missing data points of the real-time parameter set, and acquiring a plurality of missing data points;
filling missing values of a plurality of missing data points through a preset linear interpolation algorithm to obtain a filling parameter set;
performing outlier replacement on the filling parameter set to obtain a replacement parameter set;
carrying out data smoothing processing on the replacement parameter set through a preset sliding window algorithm to obtain a denoising parameter set;
carrying out correlation analysis on the denoising parameter set to obtain parameter correlation data;
and carrying out time sequence feature extraction on the denoising parameter set based on the parameter correlation data to obtain a target time sequence feature set.
3. The remote monitoring method for an appliance according to claim 1, wherein the obtaining the historical parameter set of the target appliance, performing data division on the historical parameter set to obtain a plurality of division data sets, and performing model training on a preset initial SF-ELM model through the plurality of division data sets to obtain a target SF-ELM model, includes:
Acquiring the historical parameter set, and performing data volume traversal on the historical parameter set to obtain a target data volume of the historical parameter set;
performing division range calibration on the historical parameter set through the target data volume to obtain a corresponding data division range;
performing data division on the historical parameter set through the data division range to obtain a plurality of division data sets;
performing model decomposition on the initial SF-ELM model to obtain a plurality of sub-models, and performing model weight analysis on the plurality of sub-models through a plurality of divided data sets to obtain model weight data corresponding to each sub-model;
and carrying out model integration on a plurality of sub-models through model weight data corresponding to each sub-model to obtain the target SF-ELM model.
4. The remote monitoring method for electrical appliances according to claim 1, wherein the inputting the target time sequence feature set into the target SF-ELM model for fault information prediction, obtaining initial predicted fault information, comprises:
inputting the target time sequence feature set into an input layer of the target SF-ELM model to perform data weight calculation to obtain a corresponding weight set;
Inputting the weight set into a first hidden layer of the target SF-ELM model to construct a parameter matrix, so as to obtain an initial parameter matrix;
inputting the initial parameter matrix into a second hidden layer of the target SF-ELM model to update the matrix to obtain a target electrical appliance matrix;
and inputting the target electrical appliance matrix into an output layer of the target SF-ELM model to perform data residual calculation to obtain target data residual, and generating the initial prediction fault information based on the target data residual.
5. The remote monitoring method for an electrical appliance according to claim 1, wherein the performing spatial correlation analysis on the plurality of sensors based on the initial predicted fault information to obtain spatial correlation data, and performing topology construction on the plurality of sensors based on the spatial correlation data to obtain a target topology based on the initial predicted fault information, includes:
carrying out data standardization processing on the initial prediction fault information to obtain standardized data information;
sensor pairing is carried out on the standardized data information, and target standardized data corresponding to each sensor is obtained;
Based on target standardized data corresponding to each sensor, performing covariance matrix calculation on each two sensors to obtain a plurality of covariance matrices;
carrying out space correlation analysis on a plurality of sensors through a plurality of covariance matrixes to obtain space correlation data;
based on a preset correlation threshold, extracting the significant correlation data of the spatial correlation data to obtain significant correlation data;
performing data fusion on the significant related data and the initial prediction fault data to obtain fusion related data;
and constructing topological relations of the plurality of sensors through the fusion related data to obtain the target topological relation.
6. The remote monitoring method for electrical appliances according to claim 1, wherein the data screening of the initial predicted fault information according to the target topological relation by a preset mutual information entropy screening algorithm to obtain screened fault information comprises the following steps:
extracting sensor nodes from the target topological relation to obtain information of a plurality of sensor nodes;
carrying out information association calculation on each sensor node information through a preset mutual information entropy screening algorithm to obtain information association data;
Carrying out joint probability density distribution analysis on the information association degree data to obtain a probability density distribution diagram;
performing curve development trend analysis on the probability density distribution map to obtain a corresponding curve development trend;
calibrating the data screening range of the probability density distribution map through the curve development trend to obtain at least one data screening range;
and carrying out data or o-gram selection on the initial prediction fault information through at least one data screening range to obtain screening fault information.
7. The remote monitoring method for an electrical appliance according to claim 1, wherein the performing data reconstruction on the initial predicted fault information based on the screening fault information to obtain target predicted fault information, and transmitting the target predicted fault information to a preset data processing terminal to perform remote monitoring on the target electrical appliance comprises:
performing data classification on the initial prediction fault information based on the screening fault information to obtain dominant variable data and auxiliary variable data;
constructing a data mapping relation between the dominant variable data and the auxiliary variable data to obtain a target data mapping relation;
Performing data reconstruction on the initial prediction fault information through the target data mapping relation to obtain target prediction fault information;
and transmitting the target prediction fault information to a preset data processing terminal to remotely monitor the target electric appliance.
8. A remote monitoring system for an appliance, the remote monitoring system for an appliance comprising:
the acquisition module is used for acquiring real-time parameters of the electric appliance through a plurality of sensors installed in the target electric appliance to obtain a real-time parameter set;
the extraction module is used for extracting time sequence characteristics of the real-time parameter set to obtain a target time sequence characteristic set;
the dividing module is used for acquiring a historical parameter set of the target electrical appliance, carrying out data division on the historical parameter set to obtain a plurality of divided data sets, and carrying out model training on a preset initial SF-ELM model through the plurality of divided data sets to obtain a target SF-ELM model;
the prediction module is used for inputting the target time sequence feature set into the target SF-ELM model to perform fault information prediction to obtain initial prediction fault information;
the construction module is used for carrying out space correlation analysis on the plurality of sensors based on the initial prediction fault information to obtain space correlation data, and carrying out topology relation construction on the plurality of sensors based on the space correlation data to obtain a target topology relation;
The screening module is used for carrying out data screening on the initial prediction fault information through a preset mutual information entropy screening algorithm according to the target topological relation to obtain screening fault information;
and the reconstruction module is used for carrying out data reconstruction on the initial predicted fault information based on the screening fault information to obtain target predicted fault information, and transmitting the target predicted fault information to a preset data processing terminal to carry out remote monitoring on the target electric appliance.
9. A remote monitoring device for an appliance, the remote monitoring device comprising: a memory and at least one processor, the memory having instructions stored therein;
the at least one processor invokes the instructions in the memory to cause the remote monitoring device for an appliance to perform the remote monitoring method for an appliance according to any one of claims 1-7.
10. A computer readable storage medium having instructions stored thereon, which when executed by a processor, implement the remote monitoring method for an appliance according to any of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311283691.6A CN117216481A (en) | 2023-09-28 | 2023-09-28 | Remote monitoring method and system for electric appliance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311283691.6A CN117216481A (en) | 2023-09-28 | 2023-09-28 | Remote monitoring method and system for electric appliance |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117216481A true CN117216481A (en) | 2023-12-12 |
Family
ID=89040571
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311283691.6A Pending CN117216481A (en) | 2023-09-28 | 2023-09-28 | Remote monitoring method and system for electric appliance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117216481A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117434497A (en) * | 2023-12-20 | 2024-01-23 | 深圳市宇隆移动互联网有限公司 | Indoor positioning method, device and equipment of satellite communication terminal and storage medium |
| CN117742240A (en) * | 2023-12-28 | 2024-03-22 | 广州和兴机电科技有限公司 | Remote monitoring method and system of numerical control machine tool |
| CN118013427A (en) * | 2024-04-08 | 2024-05-10 | 深圳市润诚达电力科技有限公司 | Liquid cooling heat dissipation automobile charging pile for liquid leakage early warning and early warning method |
-
2023
- 2023-09-28 CN CN202311283691.6A patent/CN117216481A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117434497A (en) * | 2023-12-20 | 2024-01-23 | 深圳市宇隆移动互联网有限公司 | Indoor positioning method, device and equipment of satellite communication terminal and storage medium |
| CN117434497B (en) * | 2023-12-20 | 2024-03-19 | 深圳市宇隆移动互联网有限公司 | Indoor positioning method, device and equipment of satellite communication terminal and storage medium |
| CN117742240A (en) * | 2023-12-28 | 2024-03-22 | 广州和兴机电科技有限公司 | Remote monitoring method and system of numerical control machine tool |
| CN118013427A (en) * | 2024-04-08 | 2024-05-10 | 深圳市润诚达电力科技有限公司 | Liquid cooling heat dissipation automobile charging pile for liquid leakage early warning and early warning method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110303380B (en) | Method for predicting residual life of cutter of numerical control machine tool | |
| CN117216481A (en) | Remote monitoring method and system for electric appliance | |
| EP2687935B1 (en) | Baseline predictive maintenance method for target device and computer program product thereof | |
| US20140365179A1 (en) | Method and Apparatus for Detecting and Identifying Faults in a Process | |
| KR101948604B1 (en) | Method and device for equipment health monitoring based on sensor clustering | |
| KR102479755B1 (en) | heavy electric equipment preventive maintenance recommendation system based on real-time sensing data and method therefor | |
| EP2978095A1 (en) | Power system operation | |
| KR102343752B1 (en) | Computer-implemented method and system for automatically monitoring and determining the status of entire process segments in a process unit | |
| CN104615122B (en) | A kind of industry control signal detection system and detection method | |
| CN117114454B (en) | DC sleeve state evaluation method and system based on Apriori algorithm | |
| CN113036913B (en) | Method and device for monitoring state of comprehensive energy equipment | |
| KR102073810B1 (en) | Method and system for predicting the failure of naval ship propulsion system using machine learning | |
| CN117114449A (en) | Visual analysis system and method for big electric power data | |
| CN117111544B (en) | Automatic-adaptation building internet of things monitoring method and system | |
| CN117639602A (en) | Self-adaptive motor running state adjusting method and system | |
| CN114896861A (en) | Rolling bearing residual life prediction method based on square root volume Kalman filtering | |
| KR101960755B1 (en) | Method and apparatus of generating unacquired power data | |
| CN117609716A (en) | Power equipment residual life remote prediction method and system based on state health information | |
| CN117408162A (en) | Power grid fault control method based on digital twin | |
| JP4170843B2 (en) | Instrument calibration support system | |
| Sim et al. | Missing data imputation algorithm for transmission systems based on multivariate imputation with principal component analysis | |
| JP7062505B2 (en) | Equipment management support system | |
| US10733533B2 (en) | Apparatus and method for screening data for kernel regression model building | |
| EP2618231A2 (en) | System and method for monitoring, diagnostics, and prognostics | |
| CN111382946B (en) | Autonomous evaluation method and system for health state of equipment and industrial internet equipment |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |