CN116306806A - Fault diagnosis model determining method and device and nonvolatile storage medium - Google Patents
Fault diagnosis model determining method and device and nonvolatile storage medium Download PDFInfo
- Publication number
- CN116306806A CN116306806A CN202310110457.7A CN202310110457A CN116306806A CN 116306806 A CN116306806 A CN 116306806A CN 202310110457 A CN202310110457 A CN 202310110457A CN 116306806 A CN116306806 A CN 116306806A
- Authority
- CN
- China
- Prior art keywords
- fault diagnosis
- diagnosis model
- time sequence
- measuring points
- sequence data
- 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 152
- 238000000034 method Methods 0.000 title claims abstract description 48
- 238000003860 storage Methods 0.000 title claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 65
- 238000005070 sampling Methods 0.000 claims abstract description 46
- 238000005259 measurement Methods 0.000 claims description 35
- 238000012360 testing method Methods 0.000 claims description 32
- 230000006870 function Effects 0.000 claims description 29
- 238000012545 processing Methods 0.000 claims description 22
- 238000011156 evaluation Methods 0.000 claims description 17
- 239000011159 matrix material Substances 0.000 claims description 7
- 238000005516 engineering process Methods 0.000 abstract description 7
- 239000000463 material Substances 0.000 abstract description 5
- 238000011176 pooling Methods 0.000 description 11
- 238000004590 computer program Methods 0.000 description 10
- 238000010586 diagram Methods 0.000 description 10
- 238000004422 calculation algorithm Methods 0.000 description 8
- 230000008569 process Effects 0.000 description 8
- 230000000694 effects Effects 0.000 description 7
- 230000004913 activation Effects 0.000 description 6
- 230000015556 catabolic process Effects 0.000 description 5
- 238000013527 convolutional neural network Methods 0.000 description 5
- 238000006731 degradation reaction Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 238000013178 mathematical model Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000009825 accumulation Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 230000008034 disappearance Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
- Y04S10/52—Outage or fault management, e.g. fault detection or location
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Test And Diagnosis Of Digital Computers (AREA)
Abstract
The invention discloses a fault diagnosis model determining method, a fault diagnosis model determining device and a nonvolatile storage medium. Wherein the method comprises the following steps: respectively sampling a plurality of measuring points in the target equipment in real time to obtain time sequence data corresponding to the plurality of measuring points, wherein the time sequence data corresponding to the plurality of measuring points are power data obtained by respectively acquiring the target equipment based on time sequence; training a preset initial fault diagnosis model based on time sequence data corresponding to the measuring points respectively to obtain a target fault diagnosis model. The invention solves the technical problems of high requirements on manpower and material resources for fault diagnosis of the power equipment and unsatisfactory accuracy of a fault diagnosis model in the related technology.
Description
Technical Field
The present invention relates to the field of electronic information technologies, and in particular, to a method and an apparatus for determining a fault diagnosis model, and a nonvolatile storage medium.
Background
Power electronics are widely used in power systems, and anomalies, failures, etc. of the power electronics can cause system operation faults, resulting in significant losses. In general, the number of components and parts of power electronic equipment is huge, the design structure and failure mechanism are complex, the traditional state monitoring means can not completely meet the monitoring and evaluation requirements, and the problems of equipment operation reliability, maintainability and the like are increasingly outstanding. The power electronic equipment fault diagnosis feature is difficult to extract, and a professional is required for fault analysis, so that a great deal of time and energy are consumed. In the related art, faults are classified by a signal processing method such as wavelet transformation, empirical mode decomposition, variational mode decomposition, and the like. Because the structure and the model of the power electronic equipment are complex, the mathematical model is not accurate enough, and the accurate fault mathematical model is difficult to build under the fault condition, so that the accuracy of fault diagnosis is not ideal, and the algorithm complexity is high.
In view of the above problems, no effective solution has been proposed at present.
Disclosure of Invention
The embodiment of the invention provides a fault diagnosis model determining method and device and a nonvolatile storage medium, which are used for at least solving the technical problems of high requirements on manpower and material resources for fault diagnosis of power equipment and non-ideal accuracy of a fault diagnosis model in the related technology.
According to an aspect of the embodiment of the present invention, there is provided a fault diagnosis model determining method including: respectively sampling a plurality of measuring points in target equipment in real time to obtain time sequence data corresponding to the plurality of measuring points, wherein the time sequence data corresponding to the plurality of measuring points are power data obtained by respectively acquiring the target equipment based on time sequence; training a preset initial fault diagnosis model based on time sequence data corresponding to the measuring points respectively to obtain a target fault diagnosis model.
According to another aspect of the embodiment of the present invention, there is provided a fault diagnosis model determination apparatus including: the sampling module is used for respectively sampling a plurality of measuring points in the target equipment in real time to obtain time sequence data corresponding to the plurality of measuring points, wherein the time sequence data corresponding to the plurality of measuring points are power data obtained by respectively acquiring the target equipment based on time sequence; the training module is used for training a preset initial fault diagnosis model based on the time sequence data corresponding to the plurality of measuring points respectively to obtain a target fault diagnosis model.
According to another aspect of the embodiments of the present invention, there is provided a nonvolatile storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform any one of the fault diagnosis model determination methods.
According to another aspect of an embodiment of the present invention, there is provided an electronic apparatus including: one or more processors and a memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the fault diagnosis model determination method of any of the above.
In the embodiment of the invention, the time sequence data corresponding to a plurality of measuring points in target equipment are obtained by respectively sampling the plurality of measuring points in real time, wherein the time sequence data corresponding to the plurality of measuring points are power data obtained by respectively acquiring the target equipment based on time sequence; training a preset initial fault diagnosis model based on time sequence data corresponding to the measuring points respectively to obtain a target fault diagnosis model. The method achieves the purposes of obtaining multidimensional time sequence data of the power equipment through time sequence data corresponding to a plurality of measuring points and improving the accuracy of a fault diagnosis model, achieves the technical effects of efficiently extracting fault characteristics and improving the accuracy of fault diagnosis, and further solves the technical problems that the requirements on manpower and material resources for fault diagnosis of the power equipment are high and the accuracy of the fault diagnosis model is not ideal in the related technology.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiments of the invention and together with the description serve to explain the invention and do not constitute a limitation on the invention. In the drawings:
FIG. 1 is a flow chart of an alternative method of determining a fault diagnostic model provided in accordance with an embodiment of the present invention;
FIG. 2 is a schematic diagram of an alternative fault diagnosis model determination method provided in accordance with an embodiment of the present invention;
fig. 3 is a schematic diagram of an alternative fault diagnosis model determining apparatus according to an embodiment of the present invention.
Detailed Description
In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
In accordance with an embodiment of the present invention, a method embodiment of fault diagnosis model determination is provided, it being noted that the steps shown in the flowchart of the figures may be performed in a computer system, such as a set of computer executable instructions, and, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in an order other than that shown or described herein.
Fig. 1 is a flowchart of a fault diagnosis model determination method according to an embodiment of the present invention, as shown in fig. 1, the method including the steps of:
step S102, respectively sampling a plurality of measuring points in the target equipment in real time to obtain time sequence data corresponding to the plurality of measuring points, wherein the time sequence data corresponding to the plurality of measuring points are power data obtained by respectively acquiring the target equipment based on time sequence.
It can be understood that the target device may be a power electronic device, and be provided with a plurality of measuring points for characterizing the operation condition of different functional modules in the target device, by sampling the plurality of measuring points in the target device in real time respectively, time sequence data corresponding to the plurality of measuring points can be obtained, where it is to be noted that the plurality of measuring points are not a plurality of points of the device in the same circuit, but are disposed in different functional modules, the target device is sampled at a plurality of angles, the time sequence data corresponding to the plurality of measuring points respectively can be regarded as multi-dimensional time sequence data of the target device, and the multi-dimensional expression of the plurality of sampling angles includes: the current angle, voltage angle, etc. may be set as desired.
Optionally, the time sequence data is one-dimensional time sequence data.
In an optional embodiment, training a preset initial fault diagnosis model based on the time sequence data corresponding to the plurality of measurement points to obtain time sequence data of a target fault diagnosis model includes: determining a sampling measuring point in the plurality of measuring points and time sequence data corresponding to the sampling measuring point; and training the initial fault diagnosis model based on the time sequence data corresponding to the sampling measuring points to obtain the target fault diagnosis model.
It can be understood that, in order to obtain a better model training effect, a plurality of measurement points are selected to obtain sampling measurement points and time sequence data corresponding to the sampling measurement points. Training the initial fault diagnosis model based on the time sequence data corresponding to the sampling measuring points to obtain a target fault diagnosis model.
Optionally, the determining manner of the sampling measurement points may be multiple, for example, a random selection manner, where under the condition that the number of the multiple measurement points is 100, multiple groups of measurement points, for example, 10 groups of 10 measurement points, are obtained according to the measurement point types or preset measurement point groups, and for each group of measurement points, 2 measurement points are randomly selected as the sampling measurement points.
In an optional embodiment, training the initial fault diagnosis model based on the time sequence data corresponding to the sampling measurement points to obtain the target fault diagnosis model includes: selecting time sequence data corresponding to the sampling measuring points by adopting a sampling window with preset duration to obtain a data selection result; and training the initial fault diagnosis model based on the data selection result to obtain the target fault diagnosis model.
It can be understood that, in order to obtain a better model training effect, a sampling window with a preset duration is adopted to select time sequence data corresponding to the sampling measurement points, in other words, time sequence data of a fixed time period of the sampling measurement points is selected, so as to obtain a data selection result. Based on the data selection result, training the initial fault diagnosis model to obtain a target fault diagnosis model.
It should be noted that, by selecting the sampling measuring point and adopting the sampling window selecting mode, the data processing mode during the selecting and selecting is realized, the data processing amount is reduced, and the training efficiency is accelerated.
Step S104, training a preset initial fault diagnosis model based on the time sequence data corresponding to the plurality of measuring points respectively to obtain a target fault diagnosis model.
It can be understood that the time sequence data corresponding to the measuring points respectively represent the fault characteristics of the target equipment from multiple angles, and the initial fault diagnosis model is trained, so that the high-accuracy target fault diagnosis model is obtained.
In an optional embodiment, training the preset initial fault diagnosis model based on the time sequence data corresponding to the plurality of measurement points to obtain a target fault diagnosis model includes: normalizing the time sequence data corresponding to the plurality of measuring points respectively to obtain the processed time sequence data corresponding to the plurality of measuring points respectively; and training the initial fault diagnosis model based on the processed time sequence data corresponding to the plurality of measuring points to obtain the target fault diagnosis model.
It can be understood that, because the time sequence data collected by the plurality of measuring points can be multiple, the corresponding data threshold values and dimensions are different, normalization processing is required to be performed on the time sequence data corresponding to the plurality of measuring points respectively, and the processed time sequence data corresponding to the plurality of measuring points respectively is obtained. Training the initial fault diagnosis model based on the processed time sequence data corresponding to the measuring points to obtain the target fault diagnosis model.
In an optional embodiment, training the preset initial fault diagnosis model based on the time sequence data corresponding to the plurality of measurement points to obtain a target fault diagnosis model includes: based on the time sequence data corresponding to the measuring points, a training set and a testing set are obtained by adopting a preset selection proportion value; training the initial fault diagnosis model by adopting the training set to obtain a candidate fault diagnosis model; and testing the candidate fault diagnosis model by adopting the test set to determine the target fault diagnosis model.
It can be understood that, based on the time sequence data corresponding to the plurality of measuring points, a preset selection proportion value is adopted, and preferably the selection proportion value can be set to be 8:2, so as to obtain a training set and a testing set. And training the initial fault diagnosis model by adopting a training set to obtain a candidate fault diagnosis model. In order to determine whether the processing capacity of the candidate fault diagnosis model meets the requirement, a test set is adopted to test the candidate fault diagnosis model, and a target fault diagnosis model is determined.
In an alternative embodiment, the testing the candidate fault diagnosis model using the test set to determine the target fault diagnosis model includes: testing the candidate fault diagnosis model by adopting the test set to obtain a test result; performing precision evaluation on the test result by adopting a preset confusion matrix to obtain a model evaluation result; and determining the target fault diagnosis model based on the model evaluation result.
It can be appreciated that the candidate fault diagnosis model is tested by adopting the test set, so as to obtain a test result. And performing precision evaluation on the test result by adopting a preset confusion matrix to obtain a model evaluation result. And determining a target fault diagnosis model based on the model evaluation result.
Alternatively, the confusion matrix may be various, for example: and adopting an N confusion matrix, wherein N is the number of evaluation indexes, and setting the number of the evaluation indexes as the dimension of the confusion matrix according to specific requirements. The evaluation index may include: evaluation indexes such as Accuracy (recorded as Accuracy), precision (recorded as Precision), recall (recorded as Recall) and F1 value.
In the following detailed description, TP represents the number of positive samples correctly evaluated, TN represents the number of negative samples correctly evaluated, FP represents the number of negative samples incorrectly evaluated, and FN represents the number of positive samples incorrectly evaluated. Accuracy is used to characterize the accuracy of model processing and can be characterized as
I.e. the ratio of the number of samples to the total number of samples, which are evaluated correctly.
The accuracy rate is used for representing the false alarm rate of model processing and can be characterized as
I.e. the ratio of the number of positive samples to the number of positive samples is evaluated correctly.
Recall is used to characterize the rate of missing messages for model processing and can be characterized as
I.e. the positive samples are correctly evaluated as the ratio of the number of positive samples to the number of actual positive samples.
The F1 value (F1-Score) is a measure for evaluating the classification ability of machine learningThe harmonic mean of the precision and recall can be characterized as
In an alternative embodiment, the initial fault diagnosis model at least includes: and under the condition of the convolution network and the residual network, training the initial fault diagnosis model by adopting the training set to obtain candidate fault diagnosis models, wherein the method comprises the following steps of: inputting the training set into the convolution network to obtain a convolved output result; processing by adopting the residual error network based on the convolved output result to obtain an initial classification result; and determining the candidate fault diagnosis model by adopting a preset loss function based on the initial classification result.
It will be appreciated that the initial fault diagnosis model includes at least: the convolution network and the residual network, wherein the convolution nerve fault diagnosis method does not need to establish an accurate mathematical model, and has better processing capacity. The residual network is favorable for solving the problems of overfitting, gradient disappearance and the like caused by the convolutional neural network, and can extract deep features. The initial fault diagnosis model is set, the advantages of the convolutional neural network and the residual error network are fully utilized, the fault characteristics are automatically and accurately extracted, and the network degradation phenomenon is avoided. Inputting the training set into a convolution network to obtain a convolved output result, processing by adopting a residual error network based on the convolved output result to obtain an initial classification result, and determining a candidate fault diagnosis model by adopting a preset loss function based on the initial classification result.
It should be noted that, as the depth of the convolutional neural network increases, a network degradation phenomenon may occur instead, resulting in a decrease in processing capacity, and a residual error module is introduced, so that the network degradation phenomenon is very skillfully eliminated by a jump connection mode, and meanwhile, the residual error network does not increase additional calculation amount and network parameters, thereby avoiding increasing algorithm complexity.
Alternatively, the initial failure diagnosis model may be set to various types, for example: the one-dimensional convolution network, the one-dimensional pooling network and the residual error network form a core network. The network parameter optimization algorithm can adopt an adagard algorithm (Adaptive Gradient, self-adaptive gradient descent algorithm), the learning rate is set to be 0.001, the initial gradient accumulation sum is 0.1, the batch number is 30, the iteration number is 100, and training of an initial fault diagnosis model is started.
Alternatively, the above-mentioned loss function may be various, for example: and (5) adopting a single-hot (one-hot) coding mode to code, and carrying out feature digitization on fault features. The multi-class cross entropy (categorical crossentropy) loss function is adopted, and is slower when the weight is reversely updated compared with the traditional mean square error loss function, and is suitable for a multi-class mode.
Optionally, the initial classification result may be output in various manners, for example: and (5) outputting an initial classification result by adopting a softmax (logistic regression) activation function.
Through the steps S102 to S104, the multi-dimensional time sequence data of the power equipment can be obtained through the time sequence data corresponding to the plurality of measuring points, so that the purpose of improving the accuracy of a fault diagnosis model is achieved, the technical effects of efficiently extracting fault characteristics and improving the accuracy of the fault diagnosis are achieved, and the technical problems that the requirements on manpower and material resources for fault diagnosis of the power equipment are high and the accuracy of the fault diagnosis model is not ideal in the related art are solved.
Based on the above examples and optional embodiments, the present invention proposes an optional embodiment, and in particular, the following description is given.
The target device may be a power electronic device, and a plurality of measurement points are provided for characterizing the operation conditions of different functional modules in the target device, the target device is sampled at a plurality of angles, and the time sequence data corresponding to the plurality of measurement points respectively may be regarded as multi-dimensional time sequence data of the target device.
In order to obtain a better model training effect, a plurality of measuring points are selected to obtain sampling measuring points and time sequence data corresponding to the sampling measuring points. And selecting time sequence data corresponding to the sampling measuring points by adopting a sampling window with preset duration to obtain a data selection result. Based on the data selection result, training the initial fault diagnosis model to obtain a target fault diagnosis model. The data processing mode during point selection is realized, the data processing amount is reduced, and the training efficiency is accelerated.
For setting and training the initial fault diagnosis model, the initial fault diagnosis model may be set to form a core network by a one-dimensional convolution network, a one-dimensional pooling network and a residual network. The network parameter optimization algorithm adopts an adagard algorithm (self-adaptive gradient descent algorithm), the learning rate is set to be 0.001, the initial gradient accumulation sum is 0.1, the batch number is 30, the iteration number is 100, the table 1 is a grid structure and parameter representation, as shown in the table 1, a layer type in an initial fault diagnosis model and the corresponding parameter setting of each layer are displayed, the Relu is an activation function commonly used in a convolutional neural network, and the "\" indicates absence. In table 1, "64@3×1" indicates that 64 represents the number of feature layers, "3*1" represents that the dimension of each feature layer is 3*1 dimensions, and similarly "2*1" indicates that the dimension of the corresponding feature layer is 2*1 dimensions.
TABLE 1
The one-dimensional convolution network is composed of kernel functions with different dimensions, the processing mode is that the kernel functions with different dimensions and the original data are subjected to discrete convolution operation, a Relu activation function is utilized to add nonlinear characteristics, the convergence speed of the convolution network is accelerated, and characteristic parameters with different dimensions of the original data are extracted. The calculation formulas are shown as (1) and (2).
X′＝f(X*W T +B) (1)
f(·)＝max(0,X*W T +B) (2)
Wherein X represents input data, which is training set input data during training, which is tested input data during testing or evaluation, X' represents the result after convolution, f (·) represents the Relu activation function, W T Representing a convolution kernel function, and B representing a bias term.
The one-dimensional pooling network is behind the convolution network, the one-dimensional data maximum pooling operation processing mode is to screen the maximum value in the sliding window as the pooling value of the current window, the downsampling of the features after the convolution operation is realized, the purpose is to accurately extract deep features, reduce model parameters, improve the calculation efficiency and reduce the fitting risk caused by large output dimension of the convolution network. The one-dimensional pooling network performs an average value or maximum value solving operation on the one-dimensional time sequence data in a pooling window to replace the data of the pooling window, wherein the maximum value operation is selected, and a formula is shown as (3).
Y out ＝MaxPool(X in ) (3)
Wherein Y is out Representing the result of maximum pooling, the MaxPool (·) function is the maximum pooling function, X in Representing the pooled data.
The residual network is used for solving the problems of overfitting, gradient disappearance and the like caused by the depth of the convolutional neural network. The traditional convolution network directly completes convolution operation from X-H (X), wherein input data X is subjected to a relation of 'H ()' function mapping, and deep features and semantic information are extracted. With the increase of the depth of the network layer, a network degradation phenomenon can appear, a residual network is introduced, the network degradation phenomenon is skillfully eliminated by a jump connection mode of X (X) +X), and meanwhile, the network is added without adding additional calculation amount and network parameters.
The time sequence data obtained by a plurality of measuring points in the embodiment can be regarded as multi-dimensional time sequence data for equipment fault diagnosis, is regarded as a classification task, and adopts a single-heat coding mode to code and digitize discrete fault characteristics. In order to overcome the defect that the traditional mean square error loss function is slow in the process of reversely updating the weight, the specific embodiment selects a multi-classification cross entropy loss function to process, the multi-classification cross entropy loss function is suitable for a multi-classification mode, a formula is shown in (4), a softmax activation function is selected for classification predicted value output, and a formula is shown in (5). The multi-classification cross entropy loss function evaluates the difference condition of the probability distribution and the real distribution of training by calculating the distance between the actual output probability and the expected output probability, the updating speed of the weight is in direct proportion to the error, when the error is large, the weight is updated quickly, and when the error is small, the weight is updated slowly.
Wherein y is i Representing the true category, a value of 1 or 0,
representing the probability of output, the range of values is (0, 1), sigma (·) represents the softmax activation function, z i Indicating the ith classification probability, categorica_loss is a multi-classification cross entropy loss function, outputsize represents the size or number of outputs, C represents the number of fault classes, and j represents an increasing subscript.
Fig. 2 is a schematic diagram of an alternative fault diagnosis model determining method according to an embodiment of the present invention, as shown in fig. 2, a target device is sampled to obtain time sequence data corresponding to a plurality of measurement points, a plurality of angles of the plurality of measurement points representing an operation condition of the target device are regarded as different dimensions, and the obtained time sequence data corresponding to the plurality of measurement points is multidimensional time sequence data. Because the data threshold values and the dimensions corresponding to the plurality of measuring points are different, normalization processing is required to be carried out on the time sequence data corresponding to the plurality of measuring points respectively, and the processed time sequence data corresponding to the plurality of measuring points respectively is obtained.
And cutting the time sequence data corresponding to the measuring points respectively according to a time window with fixed duration to obtain a training set and a testing set. Preferably, the ratio of training set to test set is 8:2. Training an initial fault diagnosis model by adopting a training set, wherein the initial fault diagnosis model comprises: one-dimensional convolution network, one-dimensional pooling network and residual error network. And obtaining candidate fault diagnosis models through multiple iterations.
And placing the test set into a candidate fault diagnosis model for testing to obtain a test result. And testing the test result by adopting the confusion matrix to obtain a model evaluation result. And determining to obtain a target fault diagnosis model based on the model evaluation result.
At least the following effects are achieved by the above alternative embodiments: aiming at the difficulty in extracting the fault diagnosis characteristics of the power electronic equipment on the engineering site, the fault analysis needs professional staff, and a great deal of time and energy are consumed. By combining the one-dimensional time sequence characteristics of the data when the power electronic equipment breaks down, the fault diagnosis problem of the power electronic equipment can be effectively solved, an accurate state evaluation model can be established, the abnormal problem during operation can be timely found and recorded, the fault diagnosis efficiency of the power electronic equipment is greatly improved, various potential safety hazards are prevented, the probability of fault outage of the power electronic equipment is reduced, the operation and maintenance costs are saved, the operation reliability of the power electronic equipment is effectively improved, the safety of a power grid is guaranteed, and the power grid has practical application value in engineering. And fully utilizes the advantages of a one-dimensional convolution network and a residual error network to automatically and accurately extract fault characteristics, thereby realizing accurate fault identification.
It should be noted that the steps illustrated in the flowcharts of the figures may be performed in a computer system such as a set of computer executable instructions, and that although a logical order is illustrated in the flowcharts, in some cases the steps illustrated or described may be performed in an order other than that illustrated herein.
The embodiment also provides a fault diagnosis model determining device, which is used for implementing the above embodiment and the preferred implementation manner, and is not described in detail. As used below, the terms "module," "apparatus" may be a combination of software and/or hardware that implements a predetermined function. While the means described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and contemplated.
According to an embodiment of the present invention, there is further provided an apparatus embodiment for implementing a fault diagnosis model determining method, and fig. 3 is a schematic diagram of a fault diagnosis model determining apparatus according to an embodiment of the present invention, as shown in fig. 3, including: the sampling module 302, training module 304, the apparatus is described below.
The sampling module 302 is configured to sample, in real time, a plurality of measurement points in a target device, to obtain time sequence data corresponding to the plurality of measurement points, where the time sequence data corresponding to the plurality of measurement points is power data obtained by the plurality of measurement points by respectively acquiring the target device based on a time sequence;
the training module 304 is connected to the sampling module 302, and is configured to train a preset initial fault diagnosis model based on the time sequence data corresponding to the plurality of measurement points, so as to obtain a target fault diagnosis model.
In the fault diagnosis model determining apparatus provided by the embodiment of the present invention, a sampling module 302 is configured to sample, in real time, a plurality of measurement points in a target device, to obtain time sequence data corresponding to the plurality of measurement points, where the time sequence data corresponding to the plurality of measurement points is power data obtained by the plurality of measurement points respectively collecting the target device based on a time sequence; the training module 304 is connected to the sampling module 302, and is configured to train a preset initial fault diagnosis model based on the time sequence data corresponding to the plurality of measurement points, so as to obtain a target fault diagnosis model. The method achieves the purposes of obtaining multidimensional time sequence data of the power equipment through time sequence data corresponding to a plurality of measuring points and improving the accuracy of a fault diagnosis model, achieves the technical effects of efficiently extracting fault characteristics and improving the accuracy of fault diagnosis, and further solves the technical problems that the requirements on manpower and material resources for fault diagnosis of the power equipment are high and the accuracy of the fault diagnosis model is not ideal in the related technology.
It should be noted that each of the above modules may be implemented by software or hardware, for example, in the latter case, it may be implemented by: the above modules may be located in the same processor; alternatively, the various modules described above may be located in different processors in any combination.
It should be noted that, the sampling module 302 and the training module 304 correspond to steps S102 to S104 in the embodiment, and the modules are the same as the examples and application scenarios implemented by the corresponding steps, but are not limited to those disclosed in the embodiment. It should be noted that the above modules may be run in a computer terminal as part of the apparatus.
It should be noted that, the optional or preferred implementation manner of this embodiment may be referred to the related description in the embodiment, and will not be repeated herein.
The fault diagnosis model determining apparatus may further include a processor and a memory, the sampling module 302, the training module 304, and the like are stored as program units in the memory, and the processor executes the program units stored in the memory to realize corresponding functions.
The processor includes a kernel, and the kernel fetches the corresponding program unit from the memory. The kernel may be provided with one or more. The memory may include volatile memory, random Access Memory (RAM), and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM), among other forms in computer readable media, the memory including at least one memory chip.
Embodiments of the present invention provide a nonvolatile storage medium having a program stored thereon, which when executed by a processor, implements a fault diagnosis model determination method.
The embodiment of the invention provides an electronic device, which comprises a processor, a memory and a program stored on the memory and capable of running on the processor, wherein the following steps are realized when the processor executes the program: respectively sampling a plurality of measuring points in target equipment in real time to obtain time sequence data corresponding to the plurality of measuring points, wherein the time sequence data corresponding to the plurality of measuring points are power data obtained by respectively acquiring the target equipment based on time sequence; training a preset initial fault diagnosis model based on the time sequence data corresponding to the plurality of measuring points respectively to obtain a target fault diagnosis model. The device herein may be a server, a PC, etc.
The invention also provides a computer program product adapted to perform, when executed on a data processing device, a program initialized with the method steps of: respectively sampling a plurality of measuring points in target equipment in real time to obtain time sequence data corresponding to the plurality of measuring points, wherein the time sequence data corresponding to the plurality of measuring points are power data obtained by respectively acquiring the target equipment based on time sequence; training a preset initial fault diagnosis model based on the time sequence data corresponding to the plurality of measuring points respectively to obtain a target fault diagnosis model.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, 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, 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.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (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 apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, 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 apparatus 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 apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, etc., such as Read Only Memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.
Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus 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 apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article or apparatus that comprises an element.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, 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, 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.
The foregoing is merely exemplary of the present invention and is not intended to limit the present invention. Various modifications and variations of the present invention will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. which come within the spirit and principles of the invention are to be included in the scope of the claims of the present invention.
Claims (10)
1. A fault diagnosis model determination method, characterized by comprising:
respectively sampling a plurality of measuring points in target equipment in real time to obtain time sequence data corresponding to the plurality of measuring points, wherein the time sequence data corresponding to the plurality of measuring points are power data obtained by respectively acquiring the target equipment based on time sequence;
training a preset initial fault diagnosis model based on time sequence data corresponding to the measuring points respectively to obtain a target fault diagnosis model.
2. The method of claim 1, wherein training a preset initial fault diagnosis model based on the time sequence data corresponding to the plurality of measuring points respectively to obtain time sequence data of a target fault diagnosis model comprises:
determining sampling measuring points in the plurality of measuring points and time sequence data corresponding to the sampling measuring points;
and training the initial fault diagnosis model based on the time sequence data corresponding to the sampling measuring points to obtain the target fault diagnosis model.
3. The method according to claim 2, wherein training the initial fault diagnosis model based on the time sequence data corresponding to the sampling measurement points to obtain the target fault diagnosis model includes:
selecting time sequence data corresponding to the sampling measuring points by adopting a sampling window with preset duration to obtain a data selection result;
and training the initial fault diagnosis model based on the data selection result to obtain the target fault diagnosis model.
4. The method of claim 1, wherein training the preset initial fault diagnosis model based on the time sequence data corresponding to the plurality of measuring points to obtain the target fault diagnosis model comprises:
based on the time sequence data corresponding to the measuring points, a training set and a testing set are obtained by adopting a preset selection proportion value;
training the initial fault diagnosis model by adopting the training set to obtain a candidate fault diagnosis model;
and testing the candidate fault diagnosis model by adopting the test set, and determining the target fault diagnosis model.
5. The method of claim 4, wherein testing the candidate fault diagnosis models using the test set to determine the target fault diagnosis model comprises:
testing the candidate fault diagnosis model by adopting the test set to obtain a test result;
performing precision evaluation on the test result by adopting a preset confusion matrix to obtain a model evaluation result;
and determining the target fault diagnosis model based on the model evaluation result.
6. The method of claim 4, wherein at least the initial fault diagnosis model includes: and under the condition of a convolution network and a residual network, training the initial fault diagnosis model by adopting the training set to obtain candidate fault diagnosis models, wherein the training set comprises the following steps:
inputting the training set into the convolution network to obtain a convolved output result;
processing by adopting the residual error network based on the convolved output result to obtain an initial classification result;
and determining the candidate fault diagnosis model by adopting a preset loss function based on the initial classification result.
7. The method according to any one of claims 1 to 6, wherein training a preset initial fault diagnosis model based on the time sequence data corresponding to each of the plurality of measurement points to obtain a target fault diagnosis model includes:
normalizing the time sequence data corresponding to the plurality of measuring points respectively to obtain processed time sequence data corresponding to the plurality of measuring points respectively;
and training the initial fault diagnosis model based on the processed time sequence data corresponding to the plurality of measuring points respectively to obtain the target fault diagnosis model.
8. A fault diagnosis model determination apparatus, characterized by comprising:
the sampling module is used for respectively sampling a plurality of measuring points in the target equipment in real time to obtain time sequence data corresponding to the plurality of measuring points, wherein the time sequence data corresponding to the plurality of measuring points are power data obtained by respectively acquiring the target equipment based on time sequence;
the training module is used for training a preset initial fault diagnosis model based on the time sequence data corresponding to the plurality of measuring points respectively to obtain a target fault diagnosis model.
9. A non-volatile storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the fault diagnosis model determination method of any one of claims 1 to 7.
10. An electronic device, comprising: one or more processors and a memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the fault diagnosis model determination method of any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310110457.7A CN116306806A (en) | 2023-02-02 | 2023-02-02 | Fault diagnosis model determining method and device and nonvolatile storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310110457.7A CN116306806A (en) | 2023-02-02 | 2023-02-02 | Fault diagnosis model determining method and device and nonvolatile storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116306806A true CN116306806A (en) | 2023-06-23 |
Family
ID=86821410
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310110457.7A Pending CN116306806A (en) | 2023-02-02 | 2023-02-02 | Fault diagnosis model determining method and device and nonvolatile storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116306806A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117081666A (en) * | 2023-09-25 | 2023-11-17 | 腾讯科技（深圳）有限公司 | Fault prediction method, device, electronic equipment, storage medium and program product |
| CN117668528A (en) * | 2024-02-01 | 2024-03-08 | 成都华泰数智科技有限公司 | Natural gas voltage regulator fault detection method and system based on Internet of things |
-
2023
- 2023-02-02 CN CN202310110457.7A patent/CN116306806A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117081666A (en) * | 2023-09-25 | 2023-11-17 | 腾讯科技（深圳）有限公司 | Fault prediction method, device, electronic equipment, storage medium and program product |
| CN117081666B (en) * | 2023-09-25 | 2024-01-09 | 腾讯科技（深圳）有限公司 | Fault prediction method, device, electronic equipment, storage medium and program product |
| CN117668528A (en) * | 2024-02-01 | 2024-03-08 | 成都华泰数智科技有限公司 | Natural gas voltage regulator fault detection method and system based on Internet of things |
| CN117668528B (en) * | 2024-02-01 | 2024-04-12 | 成都华泰数智科技有限公司 | Natural gas voltage regulator fault detection method and system based on Internet of things |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111459700B (en) | Equipment fault diagnosis method, diagnosis device, diagnosis equipment and storage medium | |
| CN116306806A (en) | Fault diagnosis model determining method and device and nonvolatile storage medium | |
| CN116450399B (en) | Fault diagnosis and root cause positioning method for micro service system | |
| CN111650453A (en) | Power equipment diagnosis method and system based on windowing characteristic Hilbert imaging | |
| CN113010389A (en) | Training method, fault prediction method, related device and equipment | |
| CN116401532B (en) | Method and system for recognizing frequency instability of power system after disturbance | |
| CN115184674A (en) | Insulation test method and device, electronic terminal and storage medium | |
| CN117034143A (en) | Distributed system fault diagnosis method and device based on machine learning | |
| CN112882898B (en) | Anomaly detection method, system, device and medium based on big data log analysis | |
| CN114548493A (en) | Method and system for predicting current overload of electric energy meter | |
| CN112380763A (en) | System and method for analyzing reliability of in-pile component based on data mining | |
| CN117170915A (en) | Data center equipment fault prediction method and device and computer equipment | |
| CN115729761B (en) | Hard disk fault prediction method, system, equipment and medium | |
| CN111930728A (en) | Method and system for predicting characteristic parameters and fault rate of equipment | |
| CN108446213A (en) | A kind of static code mass analysis method and device | |
| CN116108385A (en) | Power equipment fault type diagnosis method and system, terminal and storage medium | |
| CN115659271A (en) | Sensor abnormality detection method, model training method, system, device, and medium | |
| CN116910526A (en) | Model training method, device, communication equipment and readable storage medium | |
| CN113033845B (en) | Construction method and device for power transmission resource co-construction and sharing | |
| CN113139332A (en) | Automatic model construction method, device and equipment | |
| CN116500456B (en) | Ternary lithium battery capacity detection method based on machine learning | |
| CN114254516B (en) | Parameter probability uncertainty modeling method under deleted data | |
| CN117406007B (en) | Charging pile charging data detection method and system | |
| Verbeke et al. | Fleet-Based Remaining Useful Life Prediction of Safety-critical Electronic Devices | |
| CN118260177A (en) | AI application performance measuring method, equipment and medium based on multi-source data |
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 |