CN112508053A - Intelligent diagnosis method, device, equipment and medium based on integrated learning framework - Google Patents
Intelligent diagnosis method, device, equipment and medium based on integrated learning framework Download PDFInfo
- Publication number
- CN112508053A CN112508053A CN202011248789.4A CN202011248789A CN112508053A CN 112508053 A CN112508053 A CN 112508053A CN 202011248789 A CN202011248789 A CN 202011248789A CN 112508053 A CN112508053 A CN 112508053A
- Authority
- CN
- China
- Prior art keywords
- data
- model
- target
- fault
- training
- 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
- 238000003745 diagnosis Methods 0.000 title claims abstract description 167
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000012549 training Methods 0.000 claims abstract description 105
- 238000012795 verification Methods 0.000 claims abstract description 78
- 238000007781 pre-processing Methods 0.000 claims abstract description 32
- 238000012545 processing Methods 0.000 claims description 44
- 230000006870 function Effects 0.000 claims description 34
- 238000011156 evaluation Methods 0.000 claims description 21
- 230000002159 abnormal effect Effects 0.000 claims description 15
- 238000004088 simulation Methods 0.000 claims description 14
- 238000004590 computer program Methods 0.000 claims description 12
- 238000005070 sampling Methods 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 11
- 238000010276 construction Methods 0.000 claims description 9
- 238000004140 cleaning Methods 0.000 claims description 8
- 239000011159 matrix material Substances 0.000 claims description 8
- 238000003860 storage Methods 0.000 claims description 8
- 239000002131 composite material Substances 0.000 claims description 2
- 238000004422 calculation algorithm Methods 0.000 description 9
- 238000004364 calculation method Methods 0.000 description 7
- 230000009286 beneficial effect Effects 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000013507 mapping Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 206010063385 Intellectualisation Diseases 0.000 description 2
- 238000002788 crimping Methods 0.000 description 2
- 238000013480 data collection Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000007613 environmental effect Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
- G06F18/2148—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the process organisation or structure, e.g. boosting cascade
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D21/00—Measuring or testing not otherwise provided for
- G01D21/02—Measuring two or more variables by means not covered by a single other subclass
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/20—Ensemble learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Test And Diagnosis Of Digital Computers (AREA)
Abstract
The invention relates to an intelligent diagnosis method, device, equipment and medium based on an integrated learning framework, which comprises the following steps: collecting multi-source operation data of a target metrological verification automatic system, and preprocessing the multi-source operation data to obtain a corresponding fault data sample set; constructing a readable target data set of a preset fault diagnosis model according to the fault data sample set; the preset fault diagnosis model adopts an XGboost model, the XGboost model is trained according to the target data set, and the corresponding XGboost model after training is used as a target fault diagnosis model; and acquiring online operation data of the target metrological verification automatic system in real time, and analyzing the online operation data according to the target fault diagnosis model to predict the fault condition of the target metrological verification automatic system. The invention can accurately predict the fault condition of the metrological verification automatic system on line in real time, and improve the efficiency and accuracy of fault diagnosis.
Description
Technical Field
The invention relates to the technical field of power metering diagnosis, in particular to an intelligent diagnosis method, device, equipment and medium based on an integrated learning framework.
Background
Along with the continuous deepening of electric power information construction and the development of smart power grids and artificial intelligence technologies, the demand of intelligent metering equipment is increasing day by day, so that the metrological verification assembly tends to be automatic, standardized and streamlined, along with the popularization of intellectualization of the metrological automation verification assembly line, people rely on the verification assembly line more and more, once the verification assembly line breaks down, the verification work error and loss caused by the fact are several times or even dozens of times of labor cost, and therefore, the realization of full automation and intellectualization of metrological verification by well-made fault early warning and rapid fault positioning is guaranteed.
The current mainstream mode of the verification assembly line is to collect various system operation data in modes of various sensor technologies, video technologies, messages and the like, and then diagnose faults in a traditional data mining mode of an expert judgment method, an SDG (software development group) model and the like. In actual production, the application effect of an expert system on the fault diagnosis of a verification system is poor, on one hand, the verification system has more related equipment, the reason for the fault is not all caused by a single problem, the fault is often the combination of multiple conditions, and a fault knowledge base needs to be continuously updated and expanded; on the other hand, the fault diagnosis of the verification system mostly adopts a mode of comparing equipment measuring point data with a preset threshold, the set size of the threshold is not adjusted in real time after scientific calculation according to the change of variable factors related to the threshold, but the adjustment is determined by expert experience in most of times, and the adjustment is kept unchanged for a long time, and the objectivity is insufficient.
In view of the above-mentioned related art, the present inventors consider that there is a defect that the actual fault diagnosis is not good, and therefore, the fault diagnosis method for the metrological verification automation system still needs to be further improved.
Disclosure of Invention
In order to overcome the defects of the prior art, the application provides an intelligent diagnosis method, device, equipment and medium based on an integrated learning framework for the fault diagnosis of an electric power metrological verification automatic system, which can accurately predict the fault condition of the metrological verification automatic system in real time on line and improve the efficiency and accuracy of fault diagnosis.
The above object of the present invention is achieved by the following technical solutions:
an intelligent diagnosis method based on an ensemble learning framework, the diagnosis method comprising:
collecting multi-source operation data of a target metrological verification automatic system, and preprocessing the multi-source operation data to obtain a corresponding fault data sample set;
constructing a readable target data set of a preset fault diagnosis model according to the fault data sample set;
the preset fault diagnosis model adopts an XGboost model, the XGboost model is trained according to the target data set, and the corresponding XGboost model after training is used as a target fault diagnosis model;
and acquiring online operation data of the target metrological verification automatic system in real time, and analyzing the online operation data according to the target fault diagnosis model to predict the fault condition of the target metrological verification automatic system.
By adopting the technical scheme, the fault data sample set is obtained according to the multi-source operation data of the target metrological verification automatic system, so that initial sample data for model training is obtained, and the data types of the multi-source operation data are rich, thereby being beneficial to the training and learning of a fault diagnosis model; the fault data sample set is constructed into a target data set which can be read by the preset fault diagnosis model, so that the preset fault diagnosis model can be trained and learned according to the target data set; the XGboost model is adopted for training, so that the model training can be performed through multi-thread parallel computation, the operation speed is high, the model training is suitable for processing large-scale electric power data, the model training efficiency is further improved, the classification precision of the obtained target fault diagnosis model is high, and the fault diagnosis accuracy of the metrological verification automatic system can be effectively improved; the online operation data are obtained in real time, and the target fault diagnosis model is used for analyzing the online operation data in real time, so that the fault condition of the automatic metrological verification system is predicted in real time, and the fault classification of the automatic metrological verification system can be predicted accurately in real time in an online manner.
Optionally, the step of preprocessing the multi-source operating data to obtain a corresponding fault data sample set includes:
and carrying out data annotation on the multi-source operation data according to the historical fault condition of the target metrological verification automatic system, and taking the corresponding multi-source operation data after data annotation as a fault data sample set.
By adopting the technical scheme, the corresponding operation states are marked on the collected multi-source operation data according to the historical fault conditions of the metrological verification automatic system, fault types can be clearly distinguished according to the operation states, so that the obtained fault data sample set can be used for training a fault diagnosis model, and the accuracy of model training is improved.
Optionally, the step of constructing a target data set readable by a preset fault diagnosis model according to the fault data sample set includes:
performing data preprocessing on the fault data sample set to eliminate abnormal data, and taking the corresponding fault data sample set after data preprocessing as a target fault data sample set;
matrix construction is carried out on the multi-source operation data and target fault data sample sets to obtain readable data sets for training of preset fault diagnosis models;
and carrying out data simulation processing on the readable data set, and taking the corresponding readable data set after the data simulation processing as a target data set.
By adopting the technical scheme, abnormal data are eliminated by preprocessing the fault data sample set, so that the effectiveness of sample data is improved, and the accuracy of model training is improved; matrix construction is carried out on the multisource operation data and the target fault data sample set, so that a readable data set of a preset fault diagnosis model can be obtained; by carrying out data simulation processing on the readable data set, the problems of data imbalance and too small fault type data quantity can be reduced, so that the accuracy and generalization capability of the model are improved, and the accuracy of fault prediction of the fault diagnosis model is further improved.
Optionally, the step of performing data preprocessing on the failure data sample set to remove abnormal data includes: and carrying out data deduplication processing and data cleaning processing on the fault data sample set.
By adopting the technical scheme, repeated fields or records in the data can be removed by carrying out data deduplication on the fault data sample set, and missing values, discrete values and class characteristic digitization processing can be filled in the data, can be deleted and can be carried out by carrying out data cleaning on the fault data sample set, so that the effectiveness of the fault data sample set can be improved, and the training of a fault diagnosis model is facilitated.
Optionally, the step of training the XGBoost model according to the target data set by using the XGBoost model as the preset fault diagnosis model, and using the corresponding XGBoost model after the training as the target fault diagnosis model includes:
dividing the target data set into a training set and a test set;
inputting the training set into the XGboost model for training, returning the classification type predicted by the model by taking a preset loss function as a target, optimizing model parameters to finish the training of the XGboost model, and taking the corresponding XGboost model after the training as a target fault diagnosis model;
and inputting the test set into the target fault diagnosis model for fault diagnosis and classification, and performing performance evaluation on the target fault diagnosis model according to a preset model performance evaluation index.
By adopting the technical scheme, the XGboost model is input into the training set for training, a certain number of classification regression trees can be constructed to fit the residual error of the previous learning, and model parameters are optimized to find the optimal parameter combination of the XGboost model, so that the accuracy of fault diagnosis of the metrological verification automatic system is improved; the performance of the target fault diagnosis model is evaluated through the preset model performance evaluation index, so that the accuracy and the generalization capability of the model can be objectively evaluated, and the effectiveness of detecting the quality of the model is improved.
Optionally, the inputting the training set into the XGBoost model for training, returning a predicted classification category with a preset loss function as a target, and optimizing a model parameter to complete the training of the XGBoost model includes:
the preset loss function adopts a slightly convex loss function, regular term superposition is adopted to prevent model overfitting, and iterative training is carried out on the XGboost model;
and optimizing the model parameters by using a grid search method, wherein the optimized model parameters comprise a learning rate, the number of estimators, the maximum depth of the tree, a sample sampling rate and a column sampling rate.
By adopting the technical scheme, for the preset loss function, the difference between the real label value and the predicted label value can be measured by adopting the slightly convex loss function, the weight which can be used for smoothing the final model learning is superposed by utilizing the regular terms, the complexity of the model is punished, and overfitting is prevented, so that the target fault diagnosis model obtained by training has stronger generalization capability; the model parameters are optimized by a network search method, so that the prediction accuracy of the target fault diagnosis model obtained by optimization can be optimal, and the accuracy of fault diagnosis of the metrological verification automatic system is improved.
Optionally, the performing performance evaluation on the target fault diagnosis model by using a preset model performance evaluation index includes: the preset model performance evaluation index comprises an accuracy rate, a recall rate and a comprehensive value, wherein the accuracy rate is the ratio of the samples identified as faults to the actual samples identified as faults, the recall rate is the ratio of the samples identified as faults to the actual samples identified as faults, and the comprehensive value is determined according to the accuracy rate and the recall rate;
evaluating model performance of the target fault diagnosis model by analyzing the accuracy, recall, and composite values.
By adopting the technical scheme, the preset model performance indexes comprising the accuracy, the recall rate and the comprehensive value are set, so that the proportion of the number of samples which are actually in fault in the number of samples which are identified as in fault can be determined, the proportion of the number of samples which are identified as in fault in the number of samples which are in fault is determined, and the conditions of the two proportions are comprehensively reflected, so that the model performance can be objectively evaluated.
The second aim of the invention is realized by the following technical scheme:
an intelligent diagnostic apparatus based on an ensemble learning framework, the diagnostic apparatus comprising:
the data acquisition module is used for acquiring multi-source operation data of the target metrological verification automatic system and preprocessing the multi-source operation data to obtain a corresponding fault data sample set;
the data readable processing module is used for constructing a target data set which can be read by a preset fault diagnosis model according to the fault data sample set;
the model training module is used for training the XGboost model according to the target data set by adopting the XGboost model for the preset fault diagnosis model, and taking the corresponding XGboost model after training as a target fault diagnosis model;
and the real-time diagnosis and prediction module is used for acquiring the online operation data of the target metrological verification automatic system in real time and analyzing the online operation data according to the target fault diagnosis model so as to predict the fault condition of the target metrological verification automatic system.
By adopting the technical scheme, the data acquisition module is used for acquiring the multi-source operation data of the target metrological verification automatic system and acquiring the fault data sample set according to the multi-source operation data, so that initial sample data for model training is obtained, and the data types of the multi-source operation data are rich, thereby being beneficial to the training and learning of the fault diagnosis model; constructing the fault data sample set into a target data set which can be read by a preset fault diagnosis model through a data readable processing model, so that the preset fault diagnosis model can be trained and learned according to the target data set; the XGboost model is adopted for training through the model training module, so that the model training can be performed through multi-thread parallel computation, the operation speed is high, the model training is suitable for processing large-scale electric power data, the model training efficiency is further improved, the classification precision of the obtained target fault diagnosis model is high, and the fault diagnosis accuracy of the metrological verification automatic system can be effectively improved; the real-time diagnosis and prediction module is used for acquiring the on-line operation data of the target metrological verification automatic system in real time, and the target fault diagnosis model is used for analyzing the on-line operation data in real time, so that the fault condition of the metrological verification automatic system is predicted in real time, and the fault classification of the metrological verification automatic system can be accurately predicted in real time on line.
The third object of the invention is realized by the following technical scheme:
a computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the intelligent diagnosis method based on an ensemble learning framework when executing the computer program.
The fourth object of the invention is realized by the following technical scheme:
a computer-readable storage medium, storing a computer program which, when executed by a processor, implements the steps of the intelligent diagnosis method based on an ensemble learning framework.
In summary, the invention includes at least one of the following beneficial technical effects:
1. the XGboost model can automatically adopt multi-thread parallel computation, the operation speed is high, large-scale electric power data are suitable to be processed, the model training efficiency is further improved, the classification precision of the obtained target fault diagnosis model is high, and the fault diagnosis accuracy of the metrological verification automatic system can be effectively improved.
2. According to the method and the system, the corresponding running states are marked on the collected multi-source running data according to the historical fault conditions of the metrological verification automatic system, fault types can be clearly distinguished according to the running states, the obtained fault data sample set can be used for training a fault diagnosis model, and the accuracy of model training is improved.
3. According to the method and the device, abnormal data are eliminated by preprocessing the data of the fault data sample set, so that the effectiveness of sample data is improved, and the accuracy of model training is facilitated; matrix construction is carried out on the multisource operation data and the target fault data sample set, so that a data set which can be read by a preset fault diagnosis model can be obtained; by carrying out data simulation processing on the data set, the problems of unbalanced data and too small fault type data quantity can be reduced, so that the accuracy and generalization capability of the model are improved, and the training accuracy of the fault diagnosis model is further improved.
4. Aiming at the preset loss function for training the XGboost model, the difference between the real label value and the predicted label value can be measured by adopting the slightly convex loss function, the final learned weight of the smooth model can be superposed by utilizing the regular terms, the complexity of the model is punished, and overfitting is prevented, so that the target fault diagnosis model obtained by training has strong generalization capability.
Drawings
FIG. 1 is a flowchart of an implementation of an intelligent diagnosis method based on an ensemble learning framework according to an embodiment of the present application;
FIG. 2 is a flowchart of an implementation of step S20 of the intelligent diagnosis method based on the ensemble learning framework according to the embodiment of the present application;
FIG. 3 is a flowchart of an implementation of step S30 of the intelligent diagnosis method based on the ensemble learning framework according to the embodiment of the present application;
FIG. 4 is a flowchart of an implementation of step S32 of the intelligent diagnosis method based on the ensemble learning framework according to the embodiment of the present application;
FIG. 5 is a flowchart of an implementation of step S33 of the intelligent diagnosis method based on the ensemble learning framework according to the embodiment of the present application;
FIG. 6 is a functional block diagram of an intelligent diagnosis device based on an ensemble learning framework according to an embodiment of the present application;
FIG. 7 is a functional block diagram of a computer device according to an embodiment of the present application.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings.
The first embodiment is as follows:
in this embodiment, as shown in fig. 1, the present application discloses an intelligent diagnosis method based on an ensemble learning framework, which includes the following steps:
s10: and collecting multi-source operation data of the target metrological verification automatic system, and preprocessing the multi-source operation data to obtain a corresponding fault data sample set.
In this embodiment, the target metrological verification automatic system is a verification system to be diagnosed; the fault data sample set refers to a data sample set with a fault state in the multi-source operating data.
Specifically, the collection sources of the multi-source operating data include a target metrological verification automation line body FDB database, line body monitoring software, and operating data generated by various sensors, and in this embodiment, the multi-source operating data includes a plurality of data types. Table 1 below shows the types of multi-source operational data collected in one embodiment, for a total of 37 items of data.
TABLE 1 Multi-Source operational data types
Furthermore, preprocessing is performed on the multi-source operating data, for example, data tagging is performed on the operating state of the multi-source operating data, so that the corresponding multi-source operating data after the data tagging is used as a fault data sample set. In this embodiment, the multi-source operating data obtained by the data tagging process includes both a normal operating state and a fault operating state.
S20: and constructing a target data set which can be read by a preset fault diagnosis model according to the fault data sample set.
In this embodiment, the target data set refers to a data set which is obtained by processing a fault data sample set and can be read by a preset fault diagnosis model; the preset fault diagnosis model refers to a model to be trained and used for fault diagnosis.
Specifically, the multisource operation data and the target fault data sample set of the target metrological verification automatic system are analyzed and processed to obtain a target data set which can be read by a preset fault diagnosis model. In this embodiment, the target data set includes a model input vector and a model output vector.
S30: the preset fault diagnosis model adopts an XGboost model, the XGboost model is trained according to a target data set, and the corresponding XGboost model after the training is finished is used as a target fault diagnosis model.
In this embodiment, the target fault diagnosis model refers to a prediction model obtained after a preset fault diagnosis model is trained.
Specifically, the preset fault diagnosis model is trained by using the target data set, and the corresponding preset fault diagnosis model after training is used as the target fault diagnosis model. In this embodiment, an XGBoost model is used as a preset fault diagnosis model, and the XGBoost model is a machine learning system with extensible lifting trees, and belongs to a novel classifier based on a classification regression tree set. The XGboost supports parallelism on feature granularity, training data are sorted in advance before model training, and then the training data are stored in a block structure, so that subsequent iteration can repeatedly use the block structure, the calculation amount in the model training process is effectively reduced, and the model training efficiency is improved.
S40: and acquiring online operation data of the target metrological verification automatic system in real time, and analyzing the online operation data according to the target fault diagnosis model to predict the fault condition of the target metrological verification automatic system.
In this embodiment, the online operation data refers to operation data of the target metrological verification automatic system acquired in real time.
Specifically, online operation data are collected in real time through line body monitoring software and a sensor of the target metrological verification automatic system and then sent to a server, and the server receives the online operation data in real time.
Further, the target fault diagnosis model is used for carrying out fault diagnosis and classification on the obtained online operation data so as to predict the fault condition of the target metrological verification automatic system in real time.
In this embodiment, in step S10, that is, the step of preprocessing the multi-source operating data to obtain the corresponding fault data sample set includes:
and carrying out data annotation on the multi-source operation data according to the historical fault condition of the target metrological verification automatic system, and taking the corresponding multi-source operation data after data annotation as a fault data sample set.
Specifically, the operation and maintenance personnel label the collected multi-source operation data with corresponding operation states according to the historical fault conditions of the target metrological verification automatic system, and in a specific embodiment, the operation states of the data labels collectively include 19 states, including a normal operation state and 18 fault operation states, which are shown in table 2 below as 18 fault operation state types.
TABLE 2 faulty run State types
| Serial number | Type of failure | Serial number | Type of failure |
| 1 | Data transmission delay | 10 | Environmental impact error |
| 2 | Poor crimping result of electric energy meter | 11 | Increase in number of unqualified electric meters |
| 3 | The verification error is not in accordance with the characteristic of meter error | 12 | Track transport speed attenuation |
| 4 | Instability of standard source output power | 13 | Pressure gauge with insufficient measurement and leakage noise |
| 5 | Time-based source error drift | 14 | Slower moving speed when the conveyor belt is normal |
| 6 | Sensor transmission data anomaly | 15 | Error value inaccuracy caused by potential difference between the bits |
| 7 | Anomaly in sensor data collection | 16 | Dispatch system failure |
| 8 | RF1D reader, barcode reader failure | 17 | Standard source overload |
| 9 | The verification scheme is different from the actual verification setting | 18 | Great change of electromagnetic intensity |
In one embodiment, as shown in fig. 2, the step of constructing a target data set readable by a preset fault diagnosis model according to the fault data sample set in step S20 includes:
s21: and performing data preprocessing on the fault data sample set to eliminate abnormal data, and taking the corresponding fault data sample set after data preprocessing as a target fault data sample set.
In this embodiment, the target failure data sample set refers to a data sample set obtained by performing data preprocessing on a failure data sample set.
Specifically, data preprocessing is performed on the fault data set, in this embodiment, the data preprocessing may include data deduplication processing and data cleaning processing, and abnormal data may be cleared by performing data preprocessing on the fault data set, so that the validity of data in the fault data sample set is improved.
S22: and carrying out matrix construction on the multi-source operation data and the target fault data sample set to obtain a readable data set for training a preset fault diagnosis model.
In this embodiment, the readable data set refers to a data set required when the preset fault diagnosis model is to be subjected to model training.
Specifically, the matrix construction is carried out on multi-source operation data of the target metrological verification automatic system and an acquired target fault data sample set, and comprises the following steps: and taking the multi-source operation data as a model input vector of a preset fault diagnosis model, and taking the target fault data sample set as a model output vector of the preset fault diagnosis model, so that the model input vector and the model output vector are taken as a readable data set of the preset fault diagnosis model.
S23: and carrying out data simulation processing on the readable data set, and taking the corresponding readable data set after the data simulation processing as a target data set.
In this embodiment, the original data volume of the data generated from the line body FDB database, the line body monitoring software, and the sensor is not enough to support the accuracy of the prediction result of the training model to the required accuracy, and is not beneficial to the training of the machine learning model, so the acquired readable data set of the preset fault diagnosis model needs to be subjected to data simulation processing.
Specifically, the data simulation processing of the readable data set comprises the following steps: and carrying out data simulation processing on the readable data set by using the SMOTE algorithm, judging and generating new sample data by using the Euclidean distance method through the SMOTE algorithm, and adding the new sample data into the readable data set so as to solve the problems of data imbalance and too small fault type data quantity and further improve the accuracy and generalization capability of model diagnosis and classification.
In this embodiment, in step S21, that is, the step of performing data preprocessing on the failure data sample set to remove abnormal data includes: and carrying out data deduplication processing and data cleaning processing on the fault data sample set so as to improve the effectiveness of the sample set data.
In this embodiment, the collected multi-source operation data has duplicate records or duplicate fields, so duplicate items need to be deduplicated, where the duplicate items are first screened and judged, and in this embodiment, a sorting and merging strategy may be adopted to judge duplicate items for the obtained fault data sample set.
Specifically, data records in the fault data sample set are sorted according to a time sequence, and then whether repeated items exist in the sorted data records is detected by utilizing the similarity. In this embodiment, for the calculation of the similarity, because a large amount of standby time data exists in the production line, the euclidean distance between the sorting data and the standby time data is calculated by using the basic field matching algorithm, and if the euclidean distance is substantially 0, the euclidean distance is determined as a duplicate item, and the duplicate item is removed, and only the standby time data is retained.
Further, data cleaning is carried out on the corresponding fault data sample set after data deduplication, and the data cleaning comprises missing value processing, abnormal value processing and category characteristic digitization processing.
For missing value processing, filling of the missing value is completed by adopting a fillna method of python.
For abnormal value processing, an abnormal value refers to that one or more numerical values in sample data have larger difference compared with other numerical values, and is also called as an outlier; in this embodiment, the abnormal value replacement is performed by using a capping method, for example, a record outside a range of standard deviation three times above and below a mean value of a certain continuous variable in a failure data sample set is replaced by a standard deviation three times above and below the mean value.
For the class feature digitization processing, the class features in the metrological verification automatic system only have the characteristics of table codes, and as the table code types reflected in the table 1, the character string lengths of the table code features are different from one another, and the class feature digitization processing is performed by replacing the table code features with the character string lengths of the table code features.
In an embodiment, as shown in fig. 3, in step S30, the step of training the XGBoost model according to the target data set by using the predetermined fault diagnosis model as the XGBoost model, and using the trained XGBoost model as the target fault diagnosis model includes:
s31: the target data set is divided into a training set and a test set.
Specifically, the target data set may be divided into a training set and a test set according to a 9:1 ratio, where the training set is used to train a target fault diagnosis model meeting requirements; the test set is used to test the performance of the target fault diagnosis model.
S32: and inputting the training set into an XGboost model for training, returning the classification type predicted by the model by taking a preset loss function as a target, optimizing model parameters to finish the training of the XGboost model, and taking the corresponding XGboost model after the training as a target fault diagnosis model.
In the embodiment of the invention, a training set is input into an XGboost model, the XGboost model establishes a first classification regression tree (CART) through an approximate algorithm in a split point searching algorithm, samples of the training set are predicted according to the first class classification regression tree, predicted values and true values are compared to obtain residual errors, and the residual errors are used as new label information to construct a next classification regression tree together with the sample data to fit the residual errors. Thus, each time a tree is added, the value of the penalty function is continually reduced.
In an embodiment, as shown in fig. 4, in step S32, inputting the training set into the XGBoost model for training, returning the classification category predicted by the model with a preset loss function as a target, and optimizing the model parameters to complete the training of the XGBoost model, the method includes:
s321: the preset loss function adopts a slightly convex loss function, and the regular terms are overlapped to prevent the model from being over-fitted, so that iterative training is carried out on the XGboost model.
In particular, the training set may be represented as comprisingData set D { (x) with m-dimensional features of n samplesi,yi)}(|D|＝n,xi∈Rm,yi∈R)，xiRepresents sample data, yiRepresenting the true tag value, R is a positive integer. XGboost is a tree set model that uses a superposition of k functions to predict the output of a target.
F＝{f(x)＝wq(x)}(q:Rm→T,w∈Rm) (2)
Wherein the content of the first and second substances,
a predicted tag value represented as a predicted target output; f represents a set composed of classification regression trees (CART); q denotes sample data xiMapping to leaf nodes of the CART tree to represent the structure of a tree; t represents the number of leaf nodes in a tree; each fkThe weight is equivalent to a q mapping and the weight of a leaf node of the q mapping, and the weight is a continuous value and is beneficial to realizing an efficient optimization algorithm; w is aiRepresenting the weight of the ith leaf node. In this embodiment, the real tag value is the real fault type number of the sample data. The predicted label value is a fault type number predicted according to the training set sample data.
A preset loss function is defined as follows:
where l represents a differentiable loss function for measuring the true tag value y and the predicted tag value y
The difference between them; omega is a regular term and is specifically expressed as
System of gamma and lambda representationCounting; by means of regular term superposition of the k class tree, the weight learned by the model at last can be smoothed, the complexity of the model is punished, and overfitting is prevented. Through the preset loss function of the embodiment, a model formed by a series of relatively simple prediction functions can be finally selected, so that the model has strong generalization capability. If the regularization term is 0, then formula (2) represents a conventional gradient boosting tree.
Further, the preset loss function is optimized as follows:
wherein f istRepresenting the new tree created in the t-th iteration, f which can promote the model most is selected through formula (3)tThrough ftTo fit the predicted result of the last iteration and the residual i of the true value,
representing the predicted result of the ith sample of the t-1 th iteration.
In the process of enhancing the tree gradient, optimizing a preset loss function by using second-order Taylor expansion to simplify the function, which is specifically as follows:
wherein, the first and second guide rollers are arranged in a row,
and
which are the first and second derivatives of the predetermined loss function, respectively; i isj＝{i|q(xi) J represents the number of samples of leaf node j;
represents the optimal weight of a leaf node j of a tree given its structure q (x);
represents the quality of the tree structure q (x).
S322: and optimizing the model parameters by using a grid search method, wherein the optimized model parameters comprise a learning rate, the number of estimators, the maximum depth of the tree, a sample sampling rate and a column sampling rate.
Specifically, model parameter combinations are optimized by a grid search method to obtain the best model parameters, such as learning rate, number of estimators, maximum depth of tree, sample sampling rate, and column sampling rate.
In one embodiment, the XGBoost model is trained as follows:
the training set is input to the XGBoost model, and may be represented as D ═ X1,X2,X3,…,XdD denotes the number of samples in the training set, X1＝{x1,x2,…,xnY denotes each piece of sample data, xnThe characteristics for each dimension, y ∈ {0,1,2,3, …,18} represent the 18 fault type sequence numbers for the target metrological verification automation system as shown in table 2.
Further, the XGBoost model is trained through input of the training set by setting the XGBoost model to a preset loss function as shown in equation (5), and a prediction result is returned.
Further, optimizing the model parameter combination by a grid method, wherein the model parameter combination comprises the following parameter combinations: a learning rate (learning _ rate [ [0.075,0.1,0.125,0.15]), a number of estimators (n _ estimators [ [10,20,30,40,50]), a maximum depth of the tree (max _ depth [ [2,3,4,5,6,7,8]), a min _ child _ weight [ [1,2,3], a sample sampling rate (subsample [ [0.7,0.8,0.9]), and a column sampling rate (column _ byte [ [0.7,0.8,0.9 ]). And evaluating and optimizing the combination parameters by using the multi-classification error rate and the negative log-likelihood of the real label. The model is obtained by parameter optimization, and the model reaches the best accuracy when the parameters learning _ rate is 0.15, n _ estimators is 20, max _ depth is 4, min _ child _ weight is 1, subsampl is 0.8, and colsample _ byte is 0.8.
S33: inputting the test set into the target fault diagnosis model for fault diagnosis and classification, and performing performance evaluation on the target fault diagnosis model according to a preset model performance evaluation index, as shown in fig. 5, the step S33 specifically includes the following steps: s331: the preset model performance evaluation index comprises an accuracy rate, a recall rate and a comprehensive value, wherein the accuracy rate is the ratio of the samples identified as faults to the actual samples identified as faults, the recall rate is the ratio of the samples identified as faults to the actual samples identified as faults, and the comprehensive value is determined according to the accuracy rate and the recall rate.
The accuracy rate is the ratio of the actual fault in the number of samples identified as the fault, and the calculation formula is as follows:
wherein Q isEThe accuracy rate is represented, and TP represents the number of samples for predicting the positive type samples into the positive type; FP represents the number of samples for which negative class samples are predicted as positive class;
the recall rate is the ratio of the actual fault samples identified as faults, and the calculation formula is as follows:
wherein Q isFThe recall is indicated and FN indicates the number of samples for which positive class samples are predicted as negative class.
The comprehensive value is determined according to the accuracy rate and the recall rate, and the calculation formula is as follows:
s332: and evaluating the model performance of the target fault diagnosis model by analyzing the accuracy, the recall rate and the comprehensive value.
Specifically, the model performance of the target fault diagnosis model is evaluated by analyzing the accuracy, the recall rate and the comprehensive value.
Through verification, the overall classification performance of the target fault diagnosis model obtained in the embodiment on the 18 kinds of metrological verification automatic system fault types reaches the accuracy rate of 98% and the recall rate of 98%, and specific model performance indexes are shown in table 3. Meanwhile, the target fault diagnosis model and other machine learning algorithms of the embodiment are trained and evaluated under the condition of adopting the same sample data set, and algorithms such as limited radius nearest neighbor classification, KNN, extreme random forest, decision tree and the like are mainly compared. As shown in the model evaluation index data in table 4, it can be seen that both the accuracy and the recall index of the XGBoost model are optimal.
TABLE 3 model Performance indices
| Type of failure | Rate of accuracy | Recall rate | F1 score | Number of samples tested |
| Data transmission delay | 1.00 | 1.00 | 1.00 | 95 |
| Poor crimping result of electric energy meter | 1.00 | 1.00 | 1.00 | 179 |
| The verification error is not in accordance with the characteristic of meter error | 0.99 | 1.00 | 1.00 | 168 |
| Instability of standard source output power | 1.00 | 1.00 | 1.00 | 157 |
| Time-based source error drift | 1.00 | 1.00 | 1.00 | 231 |
| Sensor data transmission data anomalies | 1.00 | 1.00 | 1.00 | 52 |
| Anomaly in sensor data collection | 1.00 | 1.00 | 1.00 | 242 |
| Failure of RFID reader and bar code reader | 1.00 | 1.00 | 1.00 | 174 |
| The verification scheme is different from the actual verification setting | 0.95 | 0.95 | 0.95 | 77 |
| Environmental impact error | 1.00 | 1.00 | 1.00 | 200 |
| Increase in number of unqualified electric meters | 0.98 | 0.95 | 0.97 | 185 |
| Track transport speed attenuation | 0.95 | 0.97 | 0.96 | 254 |
| Pressure gauge with insufficient measurement and leakage noise | 1.00 | 1.00 | 1.00 | 132 |
| Slower moving speed when the conveyor belt is normal | 0.96 | 0.95 | 0.95 | 101 |
| Error value inaccuracy caused by potential difference between the bits | 1.00 | 1.00 | 1.00 | 158 |
| Dispatch system failure | 1.00 | 1.00 | 1.00 | 217 |
| Standard source overload | 0.98 | 0.97 | 0.97 | 204 |
| Great change of electromagnetic intensity | 1.00 | 1.00 | 1.00 | 28 |
| Data transmission delay | 1.00 | 1.00 | 1.00 | 94 |
| General of | 0.98 | 0.98 | 0.98 | 2948 |
TABLE 4 comparison of model Algorithm Performance
It should be understood that, the sequence numbers of the steps in the foregoing embodiments do not imply an execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.
Example two:
in one embodiment, an intelligent diagnosis device based on an ensemble learning framework is provided, and the intelligent diagnosis device based on the ensemble learning framework corresponds to the intelligent diagnosis method based on the ensemble learning framework in the above embodiments one to one. As shown in fig. 6, the intelligent diagnosis apparatus based on the ensemble learning framework includes a data acquisition module 10, a data readable processing module 20, a model training module 30 and a real-time diagnosis prediction module 40. The functional modules are explained in detail as follows:
the data acquisition module 10 is used for acquiring multi-source operation data of the target metrological verification automatic system and preprocessing the multi-source operation data to obtain a corresponding fault data sample set;
the data readable processing module 20 is used for constructing a target data set which can be read by a preset fault diagnosis model according to the fault data sample set;
the model training module 30 is used for presetting a fault diagnosis model, adopting an XGboost model, training the XGboost model according to a target data set, and taking the corresponding XGboost model after training as the target fault diagnosis model;
and the real-time diagnosis and prediction module 40 is used for acquiring the online operation data of the target metrological verification automatic system in real time and analyzing the online operation data according to the target fault diagnosis model so as to predict the fault condition of the target metrological verification automatic system.
Optionally, the data obtaining module 10 includes:
and the data label processing submodule is used for carrying out data annotation on the multi-source operating data according to the historical fault condition of the target metrological verification automatic system, and taking the corresponding multi-source operating data after data annotation as a fault data sample set.
Optionally, the data readable processing module 20 includes:
the data preprocessing submodule is used for preprocessing the fault data sample set to eliminate abnormal data and taking the corresponding fault data sample set after data preprocessing as a target fault data sample set;
the matrix construction submodule is used for carrying out matrix construction on the multisource operation data and the target fault data sample set so as to obtain a readable data set for training a preset fault diagnosis model;
and the simulation processing module is used for carrying out data simulation processing on the readable data set and taking the corresponding readable data set after the data simulation processing as a target data set.
Optionally, the data preprocessing sub-module includes:
and the abnormal data processing unit is used for carrying out data deduplication processing and data cleaning processing on the fault data sample set.
Optionally, the model training module 30 includes:
the data dividing submodule is used for dividing the target data set into a training set and a test set;
the model training submodule is used for inputting a training set into the XGboost model for training, returning the classification type predicted by the model by taking a preset loss function as a target, optimizing model parameters to finish the training of the XGboost model, and taking the corresponding XGboost model after the training as a target fault diagnosis model;
and the model evaluation submodule is used for inputting the test set into the target fault diagnosis model for fault diagnosis and classification and carrying out performance evaluation on the target fault diagnosis model by using a preset model performance evaluation index.
Optionally, the model training sub-module includes:
the function determining unit is used for presetting a slightly convex loss function as a loss function, overlapping a regular term to prevent the model from being over-fitted, and performing iterative training on the XGboost model;
and the parameter optimizing unit is used for optimizing the model parameters by utilizing a grid searching method, wherein the optimized model parameters comprise a learning rate, the number of estimators, the maximum depth of a tree, a sample sampling rate and a column sampling rate.
Optionally, the model evaluation sub-module includes:
the model performance evaluation index calculation unit is used for presetting a model performance evaluation index comprising an accuracy rate, a recall rate and a comprehensive value, wherein the accuracy rate is the ratio of the actual faults in the samples identified as the faults, the recall rate is the ratio of the actual faults in the samples identified as the faults, and the comprehensive value is determined according to the accuracy rate and the recall rate;
and the performance analysis unit is used for evaluating the model performance of the target fault diagnosis model by analyzing the accuracy, the recall rate and the comprehensive value.
For specific definition of the intelligent diagnosis device based on the ensemble learning framework, reference may be made to the above definition of the intelligent diagnosis method based on the ensemble learning framework, and details are not repeated here. The modules in the intelligent diagnosis device based on the integrated learning framework can be wholly or partially realized by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
Example three:
in one embodiment, a computer device is provided, which may be a server, the internal structure of which may be as shown in fig. 7. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer equipment is used for storing data such as multi-source operation data, fault data sample sets, target data sets, online operation data and other intermediate processing data. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to realize the intelligent diagnosis method based on the integrated learning framework, and the processor executes the computer program to realize the following steps:
s10: collecting multi-source operation data of a target metrological verification automatic system, and preprocessing the multi-source operation data to obtain a corresponding fault data sample set;
s20: constructing a readable target data set of a preset fault diagnosis model according to the fault data sample set;
s30: the preset fault diagnosis model adopts an XGboost model, the XGboost model is trained according to a target data set, and the corresponding XGboost model after the training is finished is used as a target fault diagnosis model;
s40: and acquiring online operation data of the target metrological verification automatic system in real time, and analyzing the online operation data according to the target fault diagnosis model to predict the fault condition of the target metrological verification automatic system.
Example four:
in one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
s10: collecting multi-source operation data of a target metrological verification automatic system, and preprocessing the multi-source operation data to obtain a corresponding fault data sample set;
s20: constructing a readable target data set of a preset fault diagnosis model according to the fault data sample set;
s30: the preset fault diagnosis model adopts an XGboost model, the XGboost model is trained according to a target data set, and the corresponding XGboost model after the training is finished is used as a target fault diagnosis model;
s40: and acquiring online operation data of the target metrological verification automatic system in real time, and analyzing the online operation data according to the target fault diagnosis model to predict the fault condition of the target metrological verification automatic system.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules to perform all or part of the above-mentioned functions.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present 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 solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments of the present invention, and are intended to be included within the scope of the present invention.
Claims (10)
1. An intelligent diagnosis method based on an ensemble learning framework, which is characterized by comprising the following steps:
collecting multi-source operation data of a target metrological verification automatic system, and preprocessing the multi-source operation data to obtain a corresponding fault data sample set;
constructing a readable target data set of a preset fault diagnosis model according to the fault data sample set;
the preset fault diagnosis model adopts an XGboost model, the XGboost model is trained according to the target data set, and the corresponding XGboost model after training is used as a target fault diagnosis model;
and acquiring online operation data of the target metrological verification automatic system in real time, and analyzing the online operation data according to the target fault diagnosis model to predict the fault condition of the target metrological verification automatic system.
2. The intelligent diagnosis method based on the ensemble learning framework according to claim 1, wherein the step of preprocessing the multi-source operation data to obtain a corresponding fault data sample set comprises:
and carrying out data annotation on the multi-source operation data according to the historical fault condition of the target metrological verification automatic system, and taking the corresponding multi-source operation data after data annotation as a fault data sample set.
3. The intelligent diagnosis method based on the ensemble learning framework according to claim 1, wherein the step of constructing a target data set readable by a preset fault diagnosis model according to the fault data sample set comprises:
performing data preprocessing on the fault data sample set to eliminate abnormal data, and taking the corresponding fault data sample set after data preprocessing as a target fault data sample set;
matrix construction is carried out on the multi-source operation data and target fault data sample sets to obtain readable data sets for training of preset fault diagnosis models;
and carrying out data simulation processing on the readable data set, and taking the corresponding readable data set after the data simulation processing as a target data set.
4. The intelligent diagnosis method based on the ensemble learning framework as claimed in claim 3, wherein the step of performing data preprocessing on the fault data sample set to remove abnormal data comprises: and carrying out data deduplication processing and data cleaning processing on the fault data sample set.
5. The intelligent diagnosis method based on the ensemble learning framework according to claim 1, wherein the preset fault diagnosis model adopts an XGBoost model, the XGBoost model is trained according to the target data set, and the step of using the corresponding XGBoost model after the training as the target fault diagnosis model comprises:
dividing the target data set into a training set and a test set;
inputting the training set into the XGboost model for training, returning the classification type predicted by the model by taking a preset loss function as a target, optimizing model parameters to finish the training of the XGboost model, and taking the corresponding XGboost model after the training as a target fault diagnosis model;
and inputting the test set into the target fault diagnosis model for fault diagnosis and classification, and performing performance evaluation on the target fault diagnosis model according to a preset model performance evaluation index.
6. The intelligent diagnosis method based on the ensemble learning framework of claim 5, wherein the inputting of the training set into the XGboost model for training, returning the predicted classification category with a preset loss function as a target, and optimizing model parameters to complete the training of the XGboost model comprises:
the preset loss function adopts a slightly convex loss function, regular term superposition is adopted to prevent model overfitting, and iterative training is carried out on the XGboost model;
and optimizing the model parameters by using a grid search method, wherein the optimized model parameters comprise a learning rate, the number of estimators, the maximum depth of the tree, a sample sampling rate and a column sampling rate.
7. The intelligent diagnosis method based on the ensemble learning framework as claimed in claim 5, wherein the performing the performance evaluation on the target fault diagnosis model by using the preset model performance evaluation index comprises:
the preset model performance evaluation index comprises an accuracy rate, a recall rate and a comprehensive value, wherein the accuracy rate is the ratio of the samples identified as faults to the actual samples identified as faults, the recall rate is the ratio of the samples identified as faults to the actual samples identified as faults, and the comprehensive value is determined according to the accuracy rate and the recall rate;
evaluating model performance of the target fault diagnosis model by analyzing the accuracy, recall, and composite values.
8. An intelligent diagnostic apparatus based on an ensemble learning framework, the diagnostic apparatus comprising:
the data acquisition module is used for acquiring multi-source operation data of the target metrological verification automatic system and preprocessing the multi-source operation data to obtain a corresponding fault data sample set;
the data readable processing module is used for constructing a target data set which can be read by a preset fault diagnosis model according to the fault data sample set;
the model training module is used for training the XGboost model according to the target data set by adopting the XGboost model for the preset fault diagnosis model, and taking the corresponding XGboost model after training as a target fault diagnosis model;
and the real-time diagnosis and prediction module is used for acquiring the online operation data of the target metrological verification automatic system in real time and analyzing the online operation data according to the target fault diagnosis model so as to predict the fault condition of the target metrological verification automatic system.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor when executing the computer program implements the steps of the intelligent diagnosis method based on an ensemble learning framework according to any one of claims 1 to 7.
10. A computer-readable storage medium, in which a computer program is stored, which, when being executed by a processor, carries out the steps of the intelligent diagnosis method based on an ensemble learning framework according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011248789.4A CN112508053A (en) | 2020-11-10 | 2020-11-10 | Intelligent diagnosis method, device, equipment and medium based on integrated learning framework |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011248789.4A CN112508053A (en) | 2020-11-10 | 2020-11-10 | Intelligent diagnosis method, device, equipment and medium based on integrated learning framework |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112508053A true CN112508053A (en) | 2021-03-16 |
Family
ID=74957626
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011248789.4A Pending CN112508053A (en) | 2020-11-10 | 2020-11-10 | Intelligent diagnosis method, device, equipment and medium based on integrated learning framework |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112508053A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113341268A (en) * | 2021-04-23 | 2021-09-03 | 国网江西省电力有限公司电力科学研究院 | Fault section positioning method utilizing magnetic field distribution characteristics below overhead line of power distribution network |
| CN113361663A (en) * | 2021-08-09 | 2021-09-07 | 国网浙江省电力有限公司金华供电公司 | Artificial intelligence-based power grid event diagnosis autonomous learning method and system |
| CN113420784A (en) * | 2021-05-28 | 2021-09-21 | 国网河北省电力有限公司营销服务中心 | Fault diagnosis system for intelligent electric meter |
| CN113469256A (en) * | 2021-07-06 | 2021-10-01 | 吉林大学重庆研究院 | Gear part mechanical damage node prediction method |
| CN113591982A (en) * | 2021-07-30 | 2021-11-02 | 山东建筑大学 | Performance evaluation method and system of fault detection and diagnosis algorithm |
| CN113791429A (en) * | 2021-08-10 | 2021-12-14 | 北京自动化控制设备研究所 | SVM-based satellite receiver fault analysis method |
| CN114638384A (en) * | 2022-05-17 | 2022-06-17 | 四川观想科技股份有限公司 | Fault diagnosis method and system based on machine learning |
| CN115289606A (en) * | 2022-07-13 | 2022-11-04 | 杭州安脉盛智能技术有限公司 | Dehumidifier online fault diagnosis method, system, server and storage medium |
| CN115456224A (en) * | 2022-11-10 | 2022-12-09 | 泽恩科技有限公司 | Intelligent operation and maintenance system of data center based on digital twins |
| CN115600695A (en) * | 2022-09-06 | 2023-01-13 | 北京航天计量测试技术研究所（Cn） | Fault diagnosis method of metering equipment |
| CN116245357A (en) * | 2023-01-31 | 2023-06-09 | 南京工大金泓能源科技有限公司 | Fault diagnosis method and system for intelligent energy-saving cabinet |
| CN117972533A (en) * | 2024-03-29 | 2024-05-03 | 北京易智时代数字科技有限公司 | Data processing method, device and equipment for industrial equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1600351A1 (en) * | 2004-04-01 | 2005-11-30 | Heuristics GmbH | Method and system for detecting defects and hazardous conditions in passing rail vehicles |
| CN106707221A (en) * | 2017-01-05 | 2017-05-24 | 云南电网有限责任公司电力科学研究院 | Fault diagnosis method and system for sensor in electric energy metering device |
| CN111444940A (en) * | 2020-02-28 | 2020-07-24 | 山东大学 | Fault diagnosis method for critical parts of fan |
| CN111832653A (en) * | 2019-07-17 | 2020-10-27 | 山东电工电气集团有限公司 | XGboost algorithm-based lightning arrester defect diagnosis method and system |
-
2020
- 2020-11-10 CN CN202011248789.4A patent/CN112508053A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1600351A1 (en) * | 2004-04-01 | 2005-11-30 | Heuristics GmbH | Method and system for detecting defects and hazardous conditions in passing rail vehicles |
| CN106707221A (en) * | 2017-01-05 | 2017-05-24 | 云南电网有限责任公司电力科学研究院 | Fault diagnosis method and system for sensor in electric energy metering device |
| CN111832653A (en) * | 2019-07-17 | 2020-10-27 | 山东电工电气集团有限公司 | XGboost algorithm-based lightning arrester defect diagnosis method and system |
| CN111444940A (en) * | 2020-02-28 | 2020-07-24 | 山东大学 | Fault diagnosis method for critical parts of fan |
Non-Patent Citations (2)
| Title |
|---|
| 房昕宇;金隼;唐水龙;: "基于S变换和XGBoost算法的进给***轴承故障诊断方法", 机械设计与研究, no. 04 * |
| 王智;吴俊盼;魏海坤;张侃健;: "基于Stacking的雷达伺服***故障诊断", 软件导刊, no. 08 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113341268B (en) * | 2021-04-23 | 2022-09-02 | 国网江西省电力有限公司电力科学研究院 | Fault section positioning method utilizing magnetic field distribution characteristics below overhead line of power distribution network |
| CN113341268A (en) * | 2021-04-23 | 2021-09-03 | 国网江西省电力有限公司电力科学研究院 | Fault section positioning method utilizing magnetic field distribution characteristics below overhead line of power distribution network |
| CN113420784A (en) * | 2021-05-28 | 2021-09-21 | 国网河北省电力有限公司营销服务中心 | Fault diagnosis system for intelligent electric meter |
| CN113469256A (en) * | 2021-07-06 | 2021-10-01 | 吉林大学重庆研究院 | Gear part mechanical damage node prediction method |
| CN113591982A (en) * | 2021-07-30 | 2021-11-02 | 山东建筑大学 | Performance evaluation method and system of fault detection and diagnosis algorithm |
| CN113361663A (en) * | 2021-08-09 | 2021-09-07 | 国网浙江省电力有限公司金华供电公司 | Artificial intelligence-based power grid event diagnosis autonomous learning method and system |
| CN113791429B (en) * | 2021-08-10 | 2024-05-07 | 北京自动化控制设备研究所 | Satellite receiver fault analysis method based on SVM |
| CN113791429A (en) * | 2021-08-10 | 2021-12-14 | 北京自动化控制设备研究所 | SVM-based satellite receiver fault analysis method |
| CN114638384A (en) * | 2022-05-17 | 2022-06-17 | 四川观想科技股份有限公司 | Fault diagnosis method and system based on machine learning |
| CN115289606A (en) * | 2022-07-13 | 2022-11-04 | 杭州安脉盛智能技术有限公司 | Dehumidifier online fault diagnosis method, system, server and storage medium |
| CN115600695A (en) * | 2022-09-06 | 2023-01-13 | 北京航天计量测试技术研究所（Cn） | Fault diagnosis method of metering equipment |
| CN115600695B (en) * | 2022-09-06 | 2023-10-17 | 北京航天计量测试技术研究所 | Fault diagnosis method for metering equipment |
| CN115456224A (en) * | 2022-11-10 | 2022-12-09 | 泽恩科技有限公司 | Intelligent operation and maintenance system of data center based on digital twins |
| CN116245357A (en) * | 2023-01-31 | 2023-06-09 | 南京工大金泓能源科技有限公司 | Fault diagnosis method and system for intelligent energy-saving cabinet |
| CN116245357B (en) * | 2023-01-31 | 2023-09-22 | 南京工大金泓能源科技有限公司 | Fault diagnosis method and system for intelligent energy-saving cabinet |
| CN117972533A (en) * | 2024-03-29 | 2024-05-03 | 北京易智时代数字科技有限公司 | Data processing method, device and equipment for industrial equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112508053A (en) | Intelligent diagnosis method, device, equipment and medium based on integrated learning framework | |
| CN110335168B (en) | Method and system for optimizing power utilization information acquisition terminal fault prediction model based on GRU | |
| JP2020500420A (en) | Yield prediction system and method for machine learning based semiconductor manufacturing | |
| CN111259947A (en) | Power system fault early warning method and system based on multi-mode learning | |
| US20230213895A1 (en) | Method for Predicting Benchmark Value of Unit Equipment Based on XGBoost Algorithm and System thereof | |
| CN111539553A (en) | Wind turbine generator fault early warning method based on SVR algorithm and skewness | |
| CN112756759B (en) | Spot welding robot workstation fault judgment method | |
| CN113298297A (en) | Wind power output power prediction method based on isolated forest and WGAN network | |
| CN116467674B (en) | Intelligent fault processing fusion updating system and method for power distribution network | |
| CN114035468B (en) | Method and system for predictively monitoring overhaul flow of fan based on XGBoost algorithm | |
| CN116316599A (en) | Intelligent electricity load prediction method | |
| CN113268883A (en) | Method for predicting corrosion rate of submarine crude oil pipeline based on PCA-ABC-SVM model | |
| CN115587672A (en) | Distribution transformer load prediction and heavy overload early warning method and system | |
| CN110781206A (en) | Method for predicting whether electric energy meter in operation fails or not by learning meter-dismantling and returning failure characteristic rule | |
| CN111898673A (en) | Dissolved oxygen content prediction method based on EMD and LSTM | |
| CN115358481A (en) | Early warning and identification method, system and device for enterprise ex-situ migration | |
| CN113449919A (en) | Power consumption prediction method and system based on feature and trend perception | |
| CN116910674A (en) | Water management monitoring method, device, equipment and medium based on data fusion inspection | |
| CN114548494A (en) | Visual cost data prediction intelligent analysis system | |
| CN116843080B (en) | Machine learning-based carbon element footprint prediction method and system for urea production | |
| Bond et al. | A hybrid learning approach to prognostics and health management applied to military ground vehicles using time-series and maintenance event data | |
| Wenqiang et al. | Remaining useful life prediction for mechanical equipment based on temporal convolutional network | |
| CN116720079A (en) | Wind driven generator fault mode identification method and system based on multi-feature fusion | |
| Chernyshev et al. | Integration of building information modeling and artificial intelligence systems to create a digital twin of the construction site | |
| CN114818460A (en) | Laboratory equipment residual service life prediction method based on automatic machine learning |
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 |