CN115130746A - Method and device for constructing fault early warning model of frequency converter and electronic equipment - Google Patents
Method and device for constructing fault early warning model of frequency converter and electronic equipment Download PDFInfo
- Publication number
- CN115130746A CN115130746A CN202210736419.8A CN202210736419A CN115130746A CN 115130746 A CN115130746 A CN 115130746A CN 202210736419 A CN202210736419 A CN 202210736419A CN 115130746 A CN115130746 A CN 115130746A
- Authority
- CN
- China
- Prior art keywords
- model
- fault
- early warning
- frequency converter
- sample 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
- 238000000034 method Methods 0.000 title claims abstract description 70
- 238000012549 training Methods 0.000 claims abstract description 62
- 238000013145 classification model Methods 0.000 claims description 17
- 238000010276 construction Methods 0.000 claims description 17
- 238000005457 optimization Methods 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 13
- 230000001419 dependent effect Effects 0.000 claims description 11
- 238000009434 installation Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 238000010219 correlation analysis Methods 0.000 claims description 4
- 238000004140 cleaning Methods 0.000 claims description 3
- 230000007613 environmental effect Effects 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 230000004224 protection Effects 0.000 abstract description 19
- 230000001105 regulatory effect Effects 0.000 abstract description 8
- 230000002159 abnormal effect Effects 0.000 description 21
- 230000008569 process Effects 0.000 description 11
- 238000005070 sampling Methods 0.000 description 11
- 238000004422 calculation algorithm Methods 0.000 description 10
- 230000005855 radiation Effects 0.000 description 10
- 238000010586 diagram Methods 0.000 description 9
- 230000006870 function Effects 0.000 description 9
- 238000012360 testing method Methods 0.000 description 9
- 238000003066 decision tree Methods 0.000 description 8
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 description 8
- 230000008859 change Effects 0.000 description 7
- 238000004519 manufacturing process Methods 0.000 description 6
- 238000012417 linear regression Methods 0.000 description 5
- 238000010200 validation analysis Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 230000000875 corresponding effect Effects 0.000 description 4
- 238000001514 detection method Methods 0.000 description 4
- 238000010438 heat treatment Methods 0.000 description 4
- 230000010365 information processing Effects 0.000 description 4
- 238000009423 ventilation Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000001816 cooling Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000010606 normalization Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 230000005856 abnormality Effects 0.000 description 2
- 238000007635 classification algorithm Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 230000020169 heat generation Effects 0.000 description 2
- 230000007774 longterm Effects 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000007476 Maximum Likelihood Methods 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 239000003990 capacitor Substances 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013499 data model Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000007636 ensemble learning method Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011425 standardization method Methods 0.000 description 1
- 238000012384 transportation and delivery Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G08—SIGNALLING
- G08B—SIGNALLING OR CALLING SYSTEMS; ORDER TELEGRAPHS; ALARM SYSTEMS
- G08B31/00—Predictive alarm systems characterised by extrapolation or other computation using updated historic data
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Artificial Intelligence (AREA)
- Emergency Management (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- Game Theory and Decision Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a method and a device for constructing a fault early warning model of a frequency converter and electronic equipment, wherein the fault early warning model comprises a fault prediction model, and the method and the device are used for constructing a first sample set based on operating data of the frequency converter; and carrying out model training based on the first sample set to obtain a fault prediction model. The fault early warning model is not based on a protection threshold value, so that the fault of the frequency converter can be early warned when the operation data are all within the range of the protection threshold value, and the safe operation of the motor speed regulating system can be ensured.
Description
Technical Field
The application relates to the technical field of fault detection, in particular to a method and a device for constructing a fault early warning model of a frequency converter and electronic equipment.
Background
The frequency converter is used as a widely used speed regulating device and plays a very important role in national production. The motor speed control system is widely applied to motor speed control systems. Once the frequency converter breaks down, the motor cannot run, and the normal production process is influenced. Especially, the sudden shutdown of the frequency converter in the process manufacturing industry will cause the interruption of the whole production process and serious loss. Therefore, in the using link, the reliability of the frequency converter has higher requirements.
Besides the failure of the frequency converter caused by the factors (design, manufacture and delivery), the use environment of the frequency converter also has a great influence on the reliable operation. In order to ensure the safety of personnel and the frequency converter, the frequency converter is provided with a plurality of threshold value protections, such as over-temperature protection, over-current protection, over-voltage protection, under-voltage protection and the like. However, in many cases, the operation data of the down converter is within the protection threshold range, but the down converter still suddenly fails. Therefore, an effective fault early warning tool is needed to early warn the fault of the frequency converter so as to ensure the safe operation of the motor speed regulating system.
Disclosure of Invention
In view of this, the present application provides a method and an apparatus for constructing a fault early warning model of a frequency converter, and an electronic device, which are used to construct a fault early warning tool capable of effectively predicting a fault of the frequency converter, so as to ensure safe operation of a motor speed regulation system.
In order to achieve the above object, the following solutions are proposed:
a method for constructing a fault early warning model of a frequency converter is applied to electronic equipment, wherein the fault early warning model comprises a fault prediction model, and the method for constructing the fault early warning model comprises the following steps:
constructing a first sample set based on the operating data of the frequency converter;
and carrying out model training based on the first sample set to obtain a fault prediction model.
Optionally, the constructing a sample set based on the historical operating data and the real-time operating data of the frequency converter includes:
collecting operation data of the frequency converter, wherein the operation data comprises historical operation data and real-time operation data;
cleaning and sorting the operation data, and integrating the cleaned and sorted operation data with the environmental information data of the installation site of the frequency converter to construct a database;
and performing correlation analysis on the data in the database, selecting independent variables and dependent variables from the data, and constructing the first sample set based on the independent variables and the dependent variables.
Optionally, the performing model training based on the first sample set to obtain a fault prediction model includes:
based on the first sample set, performing model training by adopting a technical route for comparing equipment with historical data of the equipment to obtain the fault prediction model;
based on the first sample set, performing model training by adopting a technical route for comparing equipment in the region to obtain the fault prediction model;
or based on the first sample set, performing model training by adopting a technical route compared by similar equipment to obtain the fault prediction model.
Optionally, the fault early warning model further includes an early warning period model, and the construction method further includes the steps of:
constructing a second sample set, wherein the second sample set is based on the early warning occurrence time of the fault early warning based on the fault prediction model and the actual fault occurrence time of the fault;
and carrying out model training based on the second sample set to obtain an early warning period model.
Optionally, the fault early warning model further includes a fault classification model, and the construction method further includes the steps of:
and performing model training based on an expert database to obtain the fault classification model.
Optionally, the method further comprises the steps of:
and evaluating and optimizing the fault early warning model.
Optionally, the method further comprises the steps of:
and optimizing and verifying the fault early warning model regularly.
A device for constructing a fault early warning model of a frequency converter is applied to electronic equipment, the fault early warning model comprises a fault prediction model, and the device comprises:
a first acquisition module configured to construct a first set of samples based on operational data of the frequency converter;
and the first training module is configured to perform model training based on the first sample set to obtain a fault prediction model.
Optionally, the fault early warning model further includes an early warning period model, and the constructing apparatus further includes:
the second acquisition module is configured to construct a second sample set, and the second sample set is based on the early warning occurrence time of the fault early warning based on the fault prediction model and the fault occurrence time of the actual occurrence of the fault;
and the second training module is configured to perform model training based on the second sample set to obtain an early warning period model.
Optionally, the fault early warning model further includes a fault classification model, and the constructing apparatus further includes:
and the third training module is configured to perform model training based on an expert database to obtain the fault classification model.
Optionally, the method further includes:
and the first optimization module is configured to evaluate and optimize the fault early warning model.
Optionally, the method further includes:
and the second optimization module is configured to perform optimization and verification processing on the fault early warning model periodically.
An electronic device is provided with the construction apparatus as described above.
An electronic device comprising at least one processor and a memory coupled to the processor, wherein:
the memory is for storing a computer program or instructions;
the processor is configured to execute the computer program or the instructions, so that the electronic device implements the method for constructing the fault early warning model of the frequency converter.
According to the technical scheme, the method and the device for constructing the fault early warning model of the frequency converter and the electronic equipment are disclosed, wherein the fault early warning model comprises a fault prediction model, and a first sample set is constructed by the method and the device based on the operation data of the frequency converter; and carrying out model training based on the first sample set to obtain a fault prediction model. The fault early warning model is not based on a protection threshold value, so that the fault of the frequency converter can be early warned when the operation data are all within the range of the protection threshold value, and the safe operation of the motor speed regulating system can be ensured.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
Fig. 1 is a flowchart of a method for constructing a fault early warning model of a frequency converter according to an embodiment of the present disclosure;
fig. 2 is a flowchart of a method for constructing a fault early warning model of another frequency converter according to an embodiment of the present application;
fig. 3 is a flowchart of a method for constructing a fault early warning model of a frequency converter according to another embodiment of the present disclosure;
fig. 4 is a flowchart of a method for constructing a fault early warning model of a frequency converter according to another embodiment of the present application;
fig. 5 is a flowchart of a method for constructing a fault early warning model of a frequency converter according to another embodiment of the present application;
fig. 6 is a flowchart of a method for constructing a fault early warning model of a frequency converter according to an embodiment of the present application;
fig. 7 is a flowchart of a method for constructing a fault early warning model of another frequency converter according to an embodiment of the present application;
fig. 8 is a flowchart of a method for constructing a fault early warning model of a frequency converter according to another embodiment of the present application;
fig. 9 is a flowchart of a method for constructing a fault early warning model of a frequency converter according to another embodiment of the present application;
fig. 10 is a flowchart of a method for constructing a fault early warning model of a frequency converter according to another embodiment of the present application;
fig. 11 is a block diagram of an electronic device according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be described clearly and completely with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only some embodiments of the present application, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Example one
Fig. 1 is a flowchart of a method for constructing a fault early warning model of a frequency converter according to an embodiment of the present application.
As shown in fig. 1, the construction method provided by this embodiment is applied to an electronic device, which can be understood as a computer or a server with information processing capability and data calculation capability, for constructing a fault early warning model for predicting a fault of a frequency converter. The fault early warning model at least comprises a fault prediction model, and the construction method of the fault prediction model comprises the following steps:
and S1, constructing a first sample set based on the operation data of the frequency converter.
Firstly, collecting operation data of a frequency converter, wherein the operation data comprises historical operation data and real-time operation data, and specifically comprises fault moment data and normal operation data, and the collection frequency can be set according to actual requirements, for example, the collection frequency can be set to be collected every 15 minutes; the operation data and the environment data of the frequency converter can be selected according to actual requirements, for example, the collected data can include: the method comprises the following steps of collecting time, summarizing light faults of the frequency converter, summarizing heavy faults of the frequency converter, alarming information of a power unit, inputting voltage, inputting current, outputting voltage, outputting current, given frequency, operating frequency, direct-current bus voltage, cooling fin temperature, capacitor surface temperature, temperature in a frequency converter cabinet, ambient temperature, ambient humidity, transformer temperature, reactor temperature, longitude of an installation place, dimensionality of the installation place, altitude of the installation place and the like.
Storing the collected time series working data into a database, wherein each record in the database can be stored in the following format:
then, the data is cleaned and sorted, and the correlation of each variable is calculated.
Wherein, the following cleaning method can be selected: setting the range (maximum value and minimum value) of each field, and if the real value and the predicted value exceed the range, deleting the data point; the average value or the middle value or the adjacent value of the variable in a period is adopted to assign the vacant data, for example, the period can be set as the time of one week, and the average value or the middle value of the variable of one week is used to assign the vacant data; and deleting or correcting the logically unreasonable or contradictory data through the real physical relation between the running data of the frequency converter. And after the data is cleaned and sorted, integrating the operation data and the collected installation site environment information data. And constructing a database required by data analysis.
Finally, the variables are selected and correlation analysis between the variables is performed.
The data loops can be set as dependent variables, and other variables can be set as independent variables to establish a plurality of data models. Wherein, if the failure prediction model is set to be a linear model, the Pearson correlation coefficient r can be selected p ：
Wherein, X i And Y i For different variables, i is the individual representation at a specific time point, and N is the total sampling number of the sampling period.
If it is a set fault prediction modelThe model is a nonlinear model, and a spearman grade correlation coefficient r can be selected on the basis of calculating the Pearson correlation coefficient s ：
Wherein, d i Is X i And Y i The difference between the levels after sorting of the absolute values of (a).
For example: through correlation analysis on the relevant variables of the operation data of the frequency converter, the fact that many variables of the frequency converter are strongly correlated with the temperature is found, so that the temperature is determined to be used as a dependent variable, other variables are used as independent variables, and a linear regression model can be established for calculation.
And S2, establishing a fault prediction model of the frequency converter by adopting machine learning.
Namely, a fault prediction model of the frequency converter is established by adopting a machine learning method based on the first sample data. Y ═ f (X), Y being a dependent variable, X being an independent variable. Wherein, X is the frequency converter operation data and the environmental information data of time series:
Y＝f(X 1 ,X 2 ……X n )
f (X) is a specific algorithm. The specific algorithm may optionally include classification, regression, clustering, neural network, etc. Different models may be selected according to different technical routes.
1. If the technical route of the equipment compared with the historical data of the equipment is adopted, models such as regression, ARIMA, LSTM and the like can be adopted.
When a technical route for comparing the device with the historical data of the device is adopted, a regression algorithm model is taken as an example to illustrate the model training process of the application.
Data of a certain period is collected, the temperature of the cabinet body is used as a dependent variable, other data are used as independent variables, and the data are obtained through a linear regression model:
Y＝Xβ
and calculating a value P, and comparing the calculated value with the measured value R. Wherein, Y is a column vector of a dependent variable, X is a sample matrix of an independent variable, and beta is a parameter matrix. The difference DEV, the mean square error MSE, and the goodness of fit Rsquare may be obtained. And when the difference value is higher than a preset threshold value, early warning is carried out. The linear model has the characteristics of simplicity and easy implementation, but the model easily generates multiple collinearity and autocorrelation problems.
DEV i ＝|P i -R i |
Where N is the total number of sampling cycles, i is the sampling number, and i is 1,2,3,4 … … …, N. The threshold may be designed based on historical data, for example, the threshold for the day i DEV may be 3 σ (P) i ) And when j periods exceed the threshold value continuously, judging that the frequency converter is abnormal in operation.
When the technical route of comparing the device with the historical data of the device is adopted, the ARIMA model is taken as an example to explain the model training process of the application. The ARIMA model is different from the traditional linear regression model, and has the advantages that the ARIMA model assumes that the observation value at the current stage is related to the observation value at the previous stage, so that the ARIMA model is more suitable for the real situation.
First, since variables such as temperature and humidity change periodically every day, the difference is made in the order of the number of sampling times per day. Meanwhile, since a normal device temperature should have no obvious trend, an ARIMA addition model is used, and the specific form is as follows:
wherein epsilon t ～WN(0,σ 2 )；θ(B)＝1-θ 1 B-…-θ q B q This is a moving average coefficient polynomial of order q;
this is a p-order autoregressive coefficient polynomial. Where B is the delay operator.
Then, the model is subjected to parameter estimation, and alternative estimation methods include an OLS and a maximum likelihood estimation method, specifically, the estimation process is to initialize parameters (using a moment estimation method) and then update the parameters by using an iterative method (e.g., a newton method).
Finally, whether to fail is determined based on whether the MSE and DEV exceed a threshold. Taking ARIMA (p, d, q) as an example, the general expression of ARIMA can be rewritten as x t+l ＝(ε t+l +γ 1 ∈ t+l-1 +…+γ l-1 ∈ t+1 )+(γ l ε t +γ l+1 ε t-1 + …), simple to verify, the prediction variance in phase l with minimum mean square error is: var [ e ] t (l)]＝(1+γ 1 2 +…+γ l-1 2 )σ 2 The threshold may be set according to the phase l prediction variance.
When a technical route is adopted for comparison with the historical data of the equipment, the model training can be carried out by using a lifting tree model in view of large sample size. The concrete form is as follows:
Assume that the tree model of each round of training is a CART tree.
Firstly, selecting the optimal segmentation variable j and segmentation point s, which is equivalent to solving the optimization problem
Then, the selected (j, s) is used to divide the region and determine the corresponding output value:
then, the above two steps are continuously called for the generated two sub-regions until a stop condition is satisfied, where the stop condition may be when the decrease value of the kini index is lower than a threshold set in advance.
Finally, the input space is divided into M regions, R 1 ,…,R M Generating a decision tree:
2. if a technical route is used that is compared to the devices in the area, various classification models can be used, here random forests as an example.
First, here, a forest generated by a CART tree is taken as an example. Firstly, preprocessing the characteristic, and performing PCA on the characteristic value by adopting a normalization or standardization method to avoid the phenomenon of overfitting.
Then, feature values are randomly selected, and sample values are extracted through repeated sampling to construct n CART trees. CART tree node characteristics can use the GINI index as a selection criterion,
the larger the value of the kini index, the larger the uncertainty of the sample set, which is similar to the entropy. The CART tree will select the features and cut points that reduce the GINI the most until a stop condition is reached.
3. If a technical route is used that compares with similar devices, an unsupervised learning model can be used.
The independent tree model is used for example, the independent forest is an anomaly detection algorithm, the algorithm has the characteristics of good effect and high efficiency, and the multi-high-latitude data can be effectively processed.
The independent forest is an unsupervised algorithm, a priori class labels are not needed, all attributes are not used when high-dimensional data are processed, some valuable attributes are selected through the kurtosis coefficient, then the independent forest is constructed, and the effect is excellent. First, the concept of an independent tree is given: the independent tree is a random binary tree, each node has either two child nodes or is directly a leaf node, and all attributes of D are continuous variables for a data set D.
Firstly, an independent tree needs to be constructed according to existing data, and the process of constructing an independent tree is as follows: (1) randomly selecting an attribute (2) randomly selecting a value between the maximum and minimum values of the attribute (3) sorting each record according to the retrieved value, placing less than the value to the left and more than the value to the right (4) recursively invoking the above three steps until a stop condition is met, optionally the stop condition may be that only one piece of incoming data or that the height of the tree reaches a threshold.
Then, after an independent tree is constructed based on the trained independent data, a new record can be imported for prediction. The prediction process is to search the test record from the iTree root node and determine that the prediction point falls on the leaf, and the assumption that the independent tree detects the abnormality is as follows: the outliers are generally rare and can be quickly distributed to the leaf nodes, so that in the independent tree, the outliers are generally marked as paths h (x) from the leaf nodes to the root nodes and are very short. The distance depth h (x) can therefore be used to determine whether a record belongs to an outlier. Finally, a criterion is selected to determine whether a record is an outlier.
Certainly, if the abnormal point is judged to be very hard by using one tree, an ensemble learning method can be selected, sample points and characteristics are randomly selected, a random forest of an independent tree is constructed, and then a record is transmitted to predict.
The final judgment rule is:
this function is an abnormality index used to determine whether the new input record is an abnormal value point.
b (x) e + C (t.size), where e denotes the number of edges that data x passes from the root node to the leaf node of the iTree, and t.size denotesNumber of sample points for this tree. C (n) denotes a correction value, generally,
h (n-1) can be used
This constant is estimated to be the euler constant. From the formula of the abnormal score, if the average path length of the data x in the iTrees is shorter, the score is closer to 1, which indicates that the data x is more abnormal; if the average path length of the data x in the iTrees is longer, the score is closer to 0, and the data x is more normal; if the average path length of the data x among the plurality of iTrees is close to the overall mean, the score will be around 0.5.
All the devices can be clustered by using a K-MEANS algorithm, and the devices in the category with less devices can be classified as faults, and the specific process is as follows:
first, the features are preprocessed, using normalization or normalization, in order to make the importance of each feature the same. Meanwhile, important features are given more weight, for example, temperature, humidity and the like, so as to increase the influence of the important features on the clustering result.
Secondly, randomly initializing, making t equal to 0, and randomly selecting k sample points as initial clustering centers m (0) ＝{m (0) 1 ,m (0) 2 ,…,m (0) k }. Next, clustering the samples, assigning each sample to the nearest class center, and performing iterative optimization according to a loss function, wherein the specific form of the loss function is as follows:
where C (x) is a function that partitions the clusters for each sample point,
for the distance function, a euclidean distance may be selected as the distance.
And finally, continuously calculating new class centers by using an optimization algorithm to know that iteration convergence is achieved or a stopping condition is met.
According to the technical scheme, the method for constructing the fault early warning model of the frequency converter is applied to electronic equipment, the fault early warning model comprises a fault prediction model, and the method for constructing the first sample set based on the operation data of the frequency converter is provided; and carrying out model training based on the first sample set to obtain a fault prediction model. The fault early warning model is not based on a protection threshold value, so that the fault of the frequency converter can be early warned when the operation data are all within the range of the protection threshold value, and the safe operation of the motor speed regulating system can be ensured.
In a specific implementation manner of this embodiment, the fault early-warning model further includes an early-warning period model, and based on this, the construction method further includes the following steps, as shown in fig. 2:
and S3, constructing a second sample set.
The second sample set of the present embodiment includes the warning occurrence time and the failure occurrence time at which the failure actually occurs. The early warning occurrence time is the result output by the corresponding computer system when the frequency converter is subjected to fault prediction based on the fault prediction model. And comparing the early warning occurrence time with the actual failure occurrence time to obtain an early warning period, and tagging the time when the frequency converter operation data fails from the time to the time by taking the failure occurrence time as Y.
For example, after the frequency converter is early-warned, the frequency converter is failed after 24 hours. The tag value is 24, for example, after the transducer is early-warned and the transducer is failed after 5 hours, the tag value is 5. And taking the frequency converter operation data of the early warning time as X.
| NO | Time of occurrence of early warning | Time of failure | Early warning advance (hours) |
| 1 | 5/17/2020 15:09 | 5/18/2020 15:09 | 24 |
| 2 | 5/17/2020 03:09 | 5/18/2020 08:09 | 5 |
| 3 | …… |
And S4, constructing an early warning period model based on the second sample set.
And (3) establishing an early warning period prediction model Y (F) (X), and selecting algorithms such as decision tree regression, lifting decision tree regression, Poisson regression and neural network regression.
In the following, the gradient lifting tree is taken as an example,
wherein each tree is the residual r of the predicted fault occurrence time and the actual fault occurrence time of the decision tree trained in the previous round mi I.e. r mi ＝Y i -f m-1 (X) training. Wherein m represents the tree model formed by the m-th round of training, and i is a sampling serial number.
In another specific implementation manner of this embodiment, the fault early warning model further includes a fault classification model, and based on this, the construction method further includes the following steps, as shown in fig. 3:
and S5, constructing a fault classification model based on the expert database.
By establishing an expert-oriented data warehouse, the expert is collected to classify possible fault reasons, such as self heating increase (abnormal loss increase) of a frequency converter, short circuit heating, faults of a cooling fan, abnormal environment temperature and the like, and a fault classification model is established by utilizing a classification algorithm. For example, a linear regression model may be selected, and when the difference between the predicted value and the theoretical value of the selected feature by the model exceeds a threshold, an alarm is given, for example, as follows:
and (3) a temperature model, wherein after the fault prediction model gives an alarm, the operation condition of the frequency converter is changed, and factors causing the change may be as follows: the frequency converter cabinet top heat radiation fan is abnormal, the filter screen is blocked, the closed environment of the frequency converter chamber is damaged (for example, doors and windows are not closed), the ventilation and heat radiation equipment such as an air conditioner is abnormal, and the self heat radiation of the frequency converter is increased.
And (3) humidity model prediction, when the fault prediction model gives an alarm, the operation condition of the frequency converter is changed, and the factors causing the change may be as follows: the frequency converter cabinet top heat radiation fan is abnormal, the filter screen is blocked, the closed environment of the frequency converter chamber is damaged (for example, doors and windows are not closed), the ventilation and heat radiation equipment such as an air conditioner is abnormal, the self heat generation of the frequency converter is increased, the environment humidity is abnormal, and the like.
In another specific implementation manner of this embodiment, the method further includes the following steps, as shown in fig. 4:
and S6, training the fault early warning model and adjusting and optimizing parameters.
Decomposing historical data of a period of time into a training set and a test set, establishing a regression model by using the training set, evaluating the quality of the model by using the test set, and evaluating the model.
The model was evaluated, with the following evaluation criteria being selected:
linear models can be evaluated with Rsquare,
wherein,
is a predicted value of the ith time point,
is the average of the dependent variable over a sampling period.
Drawing a comparison graph of the real value and the model predicted value, and judging whether an abnormal detection rule can be established by an expert, for example, the following method can be selected:
historical training set, test set and full set actual value, predicted value comparison graph and residual graph (used for evaluating model fitting degree);
the historical training set, the test set and the full set of actual values, the predicted value comparison graph and the residual graph (used for reflecting the long-term trend of the data after short-term fluctuation removal) are smoothed by a time window (the time window is optional);
after the model is built, training and verification work is required to optimize the model. Wherein the validation optimization comprises comparing the model predictive data using training set validation of historical data and real-time data.
In another specific implementation manner of this embodiment, the method further includes the following steps, as shown in fig. 5:
and S7, periodically optimizing and verifying the fault early warning model.
The accuracy of the model in the last period of time (such as one month) can be returned during each operation of the model, and the system can judge whether the model needs to be adjusted or not according to the change trend of the accuracy (if the accuracy is greatly reduced in the near term, the model needs to be retrained).
For example, the number of first-type and second-type alarms in the last period of time may be counted periodically (for example, if there is a label feedback, the real alarm, the false alarm frequency, etc. may be counted), and the system may determine whether the model needs to be adjusted according to the variation trend of the accuracy.
Example two
Fig. 6 is a block diagram of a device for constructing a fault early warning model of a frequency converter according to an embodiment of the present application.
As shown in fig. 6, the construction method provided by this embodiment is applied to an electronic device for constructing a fault early warning model for predicting a fault of a frequency converter, and the electronic device may be understood as a computer or a server having information processing capability and data computing capability. The fault early warning model at least comprises a fault prediction model, and the construction device of the fault prediction model comprises a first acquisition module 10 and a first training module 20.
The first acquisition module is used for constructing a first sample set based on the operation data of the frequency converter. The means of constructing the first sample set of this module is the same as step S1 in the first embodiment, and is not described here again.
The first training module is used for establishing a fault prediction model of the frequency converter by adopting machine learning. The means of establishing the fault prediction model in this module is the same as step S2 in the first embodiment, and is not described here again.
According to the technical scheme, the device for constructing the fault early warning model of the frequency converter is applied to electronic equipment, the fault early warning model comprises a fault prediction model, and the constructing device is used for constructing a first sample set based on the operation data of the frequency converter; and carrying out model training based on the first sample set to obtain a fault prediction model. The fault early warning model is not based on a protection threshold value, so that the fault of the frequency converter can be early warned when the operation data are all within the range of the protection threshold value, and the safe operation of the motor speed regulating system can be ensured.
In a specific implementation manner of this embodiment, the fault early warning model further includes an early warning period model, and based on this, the construction apparatus further includes a second acquisition module 30 and a second training module 40, as shown in fig. 7:
the second acquisition module is used to construct a second sample set.
The second sample set of the present embodiment includes the warning occurrence time and the failure occurrence time at which the failure actually occurs. The early warning occurrence time is the result output by the corresponding computer system when the frequency converter is subjected to fault prediction based on the fault prediction model. And comparing the early warning occurrence time with the actual failure occurrence time to obtain an early warning period, and tagging the time of failure occurrence of the frequency converter operation data from the time distance by taking the failure occurrence time as Y.
For example, after the frequency converter is early-warned, the frequency converter is failed after 24 hours. The label value is 24, for example, after the frequency converter is early warned, and after 5 hours, the frequency converter is failed, the label value is 5. And taking the frequency converter operation data of the early warning time as X.
The second training module is used for constructing an early warning period model based on the second sample set.
And (3) establishing an early warning period prediction model Y (F) (X), and selecting algorithms such as decision tree regression, lifting decision tree regression, Poisson regression and neural network regression.
In the following, the gradient lifting tree is taken as an example,
wherein each tree is the residual r of the predicted fault occurrence time and the actual fault occurrence time of the decision tree trained in the previous round min I.e. r min ＝Y i -f m-1 (X) training. Wherein m represents the tree model formed by the m-th round of training, and i is a sampling serial number.
In a further specific implementation manner of this embodiment, the fault early warning model further includes a fault classification model, and based on this, the constructing apparatus further includes a third training module 50, as shown in fig. 8:
and the third training module is used for constructing a fault classification model based on the expert database.
By establishing an expert-oriented data warehouse, the expert is collected to classify possible fault reasons, such as self heating increase (abnormal loss increase) of a frequency converter, short circuit heating, faults of a cooling fan, abnormal environment temperature and the like, and a fault classification model is established by utilizing a classification algorithm. For example, a linear regression model may be selected, and when the difference between the predicted value and the theoretical value of the selected feature by the model exceeds a threshold, an alarm is given, for example, as follows:
and (3) a temperature model, wherein after the fault prediction model gives an alarm, the operation condition of the frequency converter is changed, and factors causing the change may be as follows: the frequency converter cabinet top heat radiation fan is abnormal, the filter screen is blocked, the closed environment of the frequency converter chamber is damaged (for example, doors and windows are not closed), the ventilation and heat radiation equipment such as an air conditioner is abnormal, and the self heat radiation of the frequency converter is increased.
And (3) humidity model prediction, when the fault prediction model gives an alarm, the operation condition of the frequency converter is changed, and factors causing the change may be as follows: the frequency converter cabinet top heat radiation fan is abnormal, the filter screen is blocked, the closed environment of the frequency converter chamber is damaged (for example, doors and windows are not closed), the ventilation and heat radiation equipment such as an air conditioner is abnormal, the self heat generation of the frequency converter is increased, the environment humidity is abnormal, and the like.
In another specific implementation manner of this embodiment, the method further includes a first optimization module 60, as shown in fig. 9:
the first optimization module is used for training the fault early warning model and adjusting and optimizing parameters.
Decomposing historical data of a period of time into a training set and a testing set, establishing a regression model by using the training set, evaluating the quality of the model by using the testing set, and evaluating the model.
The model was evaluated, with the following optional evaluation criteria:
linear models can be evaluated with Rsquare,
wherein,
is a predicted value of the ith time point,
is the average of the dependent variable over a sampling period.
Drawing a comparison graph of the real value and the model predicted value, and judging whether an abnormal detection rule can be established by an expert, for example, the following method can be selected:
historical training set, test set and full set actual value, predicted value comparison graph and residual graph (used for evaluating model fitting degree);
the historical training set, the test set and the full set of actual values, the predicted value comparison graph and the residual graph (used for reflecting the long-term trend of the data after short-term fluctuation removal) are smoothed by a time window (the time window is optional);
after the model is built, training and validation work is required to optimize the model. Wherein the validation optimization comprises comparing the model predictive data using training set validation of historical data and real-time data.
In another specific implementation manner of this embodiment, a second optimization module 70 is further included, as shown in fig. 10:
the second optimization module is used for periodically optimizing and verifying the fault early warning model.
The accuracy of the model in the last period of time (such as one month) can be returned during each operation of the model, and the system can judge whether the model needs to be adjusted or not according to the change trend of the accuracy (if the accuracy is greatly reduced in the near term, the model needs to be retrained).
For example, the number of first-class and second-class alarms in the last period of time may be periodically counted (for example, one month, if there is label feedback, the real alarm, the false alarm frequency, etc. may be counted), and the system may determine whether the model needs to be adjusted according to the variation trend of the accuracy.
EXAMPLE III
The present embodiment provides an electronic device that can be understood as a computer or server having data calculation capability and information processing capability. The electronic equipment is provided with the method for constructing the fault early warning model of the frequency converter, which is provided by the previous embodiment, the method is applied to the electronic equipment, the fault early warning model comprises a fault prediction model, and the constructing device is used for constructing a first sample set based on the operation data of the frequency converter; and carrying out model training based on the first sample set to obtain a fault prediction model. The fault early warning model is not based on a protection threshold value, so that the fault of the frequency converter can be early warned when the operation data are all in the range of the protection threshold value, and the safe operation of the motor speed regulating system can be ensured. In addition, the fault early warning model also comprises a fault period model and a fault classification model.
Example four
Fig. 11 is a block diagram of an electronic device according to an embodiment of the present application.
As shown in fig. 11, the electronic device provided in the present embodiment can be understood as a computer or a server having data calculation capability and information processing capability. The electronic device comprises at least one processor 101 and a memory 102, which are connected by a data bus 103. The memory is used for storing a computer program or instructions, and the processor is used for executing the corresponding computer program or instructions to enable the electronic device to implement a method for constructing a fault early warning model of a frequency converter in the first embodiment, wherein the fault early warning model comprises a fault prediction model, and the method for constructing the first sample set is specifically based on the operating data of the frequency converter; and carrying out model training based on the first sample set to obtain a fault prediction model. The fault early warning model is not based on a protection threshold value, so that the fault of the frequency converter can be early warned when the operation data are all within the range of the protection threshold value, and the safe operation of the motor speed regulating system can be ensured. In addition, the fault early warning model also comprises a fault period model and a fault classification model.
The embodiments in the present specification are all described in a progressive manner, and each embodiment focuses on differences from other embodiments, and portions that are the same and similar between the embodiments may be referred to each other.
As will be appreciated by one of skill in the art, embodiments of the present invention may be provided as a method, apparatus, or computer program product. Accordingly, embodiments of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, embodiments of the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
Embodiments of the present invention are described with reference to flowchart illustrations and/or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing terminal to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing terminal to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing terminal to cause a series of operational steps to be performed on the computer or other programmable terminal to produce a computer implemented process such that the instructions which execute on the computer or other programmable terminal provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications of these embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the embodiments of the invention.
Finally, it should also be noted that, in this document, relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or terminal that comprises the element.
The technical solutions provided by the present invention are described in detail above, and the principle and the implementation of the present invention are explained in this document by applying specific examples, and the descriptions of the above examples are only used to help understanding the method and the core idea of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (14)
1. A method for constructing a fault early warning model of a frequency converter is applied to electronic equipment and is characterized in that the fault early warning model comprises a fault prediction model, and the method comprises the following steps:
constructing a first sample set based on the operating data of the frequency converter;
and carrying out model training based on the first sample set to obtain a fault prediction model.
2. The method of constructing as claimed in claim 1, wherein said constructing a sample set based on historical operating data and real-time operating data of said frequency converter comprises the steps of:
collecting operation data of the frequency converter, wherein the operation data comprises historical operation data and real-time operation data;
cleaning and sorting the operation data, and integrating the cleaned and sorted operation data with the environmental information data of the installation site of the frequency converter to construct a database;
and carrying out correlation analysis on the data in the database, selecting independent variables and dependent variables from the data, and constructing the first sample set based on the independent variables and the dependent variables.
3. The construction method according to claim 1, wherein the model training based on the first sample set to obtain a fault prediction model comprises the steps of:
based on the first sample set, performing model training by adopting a technical route for comparing equipment with historical data of the equipment to obtain the fault prediction model;
based on the first sample set, performing model training by adopting a technical route for comparing equipment in the region to obtain the fault prediction model;
or based on the first sample set, performing model training by adopting a technical route compared by similar equipment to obtain the fault prediction model.
4. The method of construction of claim 1, wherein the fault-early-warning model further comprises an early-warning period model, the method of construction further comprising the steps of:
constructing a second sample set, wherein the second sample set is based on the early warning occurrence time of the fault early warning based on the fault prediction model and the actual fault occurrence time of the fault;
and carrying out model training based on the second sample set to obtain an early warning period model.
5. The method of construction of claim 1, wherein the fault early warning model further comprises a fault classification model, the method of construction further comprising the steps of:
and carrying out model training based on an expert database to obtain the fault classification model.
6. The construction method according to any one of claims 1 to 5, further comprising the steps of:
and evaluating and optimizing the fault early warning model.
7. The method of constructing as claimed in claim 6, further comprising the steps of:
and optimizing and verifying the fault early warning model regularly.
8. The utility model provides a device for building fault early warning model of converter, is applied to electronic equipment, its characterized in that, fault early warning model includes the fault prediction model, the device for building includes:
a first acquisition module configured to construct a first set of samples based on operational data of the frequency converter;
and the first training module is configured to perform model training based on the first sample set to obtain a fault prediction model.
9. The build device of claim 8, wherein the fault-warning model further comprises a warning period model, the build device further comprising:
the second acquisition module is configured to construct a second sample set, and the second sample set is based on the early warning occurrence time of the fault early warning based on the fault prediction model and the fault occurrence time of the actual occurrence of the fault;
and the second training module is configured to perform model training based on the second sample set to obtain an early warning period model.
10. The build device of claim 8, wherein the fault-early-warning model further comprises a fault classification model, the build device further comprising:
and the third training module is configured to perform model training based on an expert database to obtain the fault classification model.
11. The building apparatus according to any one of claims 8 to 10, further comprising:
and the first optimization module is configured to evaluate and optimize the fault early warning model.
12. The build device of claim 12, further comprising:
and the second optimization module is configured to perform optimization and verification processing on the fault early warning model periodically.
13. An electronic device, characterized in that a construction apparatus according to any one of claims 8 to 12 is provided.
14. An electronic device comprising at least one processor and a memory coupled to the processor, wherein:
the memory is used for storing computer programs or instructions;
the processor is used for executing the computer program or the instructions to enable the electronic equipment to realize the method for constructing the fault early warning model of the frequency converter according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210736419.8A CN115130746A (en) | 2022-06-27 | 2022-06-27 | Method and device for constructing fault early warning model of frequency converter and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210736419.8A CN115130746A (en) | 2022-06-27 | 2022-06-27 | Method and device for constructing fault early warning model of frequency converter and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115130746A true CN115130746A (en) | 2022-09-30 |
Family
ID=83380258
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210736419.8A Pending CN115130746A (en) | 2022-06-27 | 2022-06-27 | Method and device for constructing fault early warning model of frequency converter and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115130746A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118246905A (en) * | 2024-05-23 | 2024-06-25 | 北京金泰康辰生物科技有限公司 | Small molecule detection equipment maintenance management system based on data analysis |
-
2022
- 2022-06-27 CN CN202210736419.8A patent/CN115130746A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118246905A (en) * | 2024-05-23 | 2024-06-25 | 北京金泰康辰生物科技有限公司 | Small molecule detection equipment maintenance management system based on data analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108762228B (en) | Distributed PCA-based multi-working-condition fault monitoring method | |
| CN111708343B (en) | Method for detecting abnormal behavior of field process behavior in manufacturing industry | |
| KR102215690B1 (en) | Method and apparatus for time series data monitoring | |
| CN112862012A (en) | Operation and maintenance system abnormity early warning method, device and equipment based on LSTM model | |
| CN113378990B (en) | Flow data anomaly detection method based on deep learning | |
| CN110571792A (en) | Analysis and evaluation method and system for operation state of power grid regulation and control system | |
| JP2002318617A (en) | Management method for manufacture process | |
| Wang et al. | Fault detection and diagnosis for multiple faults of VAV terminals using self-adaptive model and layered random forest | |
| CN113657814B (en) | Aviation network risk prediction method and risk grade evaluation method | |
| CN111625516A (en) | Method and device for detecting data state, computer equipment and storage medium | |
| CN113935440A (en) | Iterative evaluation method and system for error state of voltage transformer | |
| CN113283657B (en) | Prediction method for electric power emergency material demand | |
| CN116737510B (en) | Data analysis-based intelligent keyboard monitoring method and system | |
| CN113408548A (en) | Transformer abnormal data detection method and device, computer equipment and storage medium | |
| CN115130746A (en) | Method and device for constructing fault early warning model of frequency converter and electronic equipment | |
| CN111680712B (en) | Method, device and system for predicting oil temperature of transformer based on similar time in day | |
| CN113884807B (en) | Power distribution network fault prediction method based on random forest and multi-layer architecture clustering | |
| US20230297095A1 (en) | Monitoring device and method for detecting anomalies | |
| CN110781206A (en) | Method for predicting whether electric energy meter in operation fails or not by learning meter-dismantling and returning failure characteristic rule | |
| CN108446202B (en) | Method for judging safety condition of machine room equipment | |
| CN111709437B (en) | Abnormal behavior detection method oriented to field process behavior of petrochemical industry | |
| CN117113233A (en) | Hierarchical energy structure scene construction method and energy consumption abnormal link tracing method | |
| CN117251811A (en) | Method, device, computer equipment and storage medium for detecting abnormal data | |
| CN111461565A (en) | Power supply side power generation performance evaluation method under power regulation | |
| CN116151799A (en) | BP neural network-based distribution line multi-working-condition fault rate rapid assessment method |
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 |