CN114358085B - Line fault diagnosis method and device based on heterogeneous model fusion - Google Patents
Line fault diagnosis method and device based on heterogeneous model fusion Download PDFInfo
- Publication number
- CN114358085B CN114358085B CN202210020481.7A CN202210020481A CN114358085B CN 114358085 B CN114358085 B CN 114358085B CN 202210020481 A CN202210020481 A CN 202210020481A CN 114358085 B CN114358085 B CN 114358085B
- Authority
- CN
- China
- Prior art keywords
- fault
- diagnosis
- traveling wave
- line fault
- features
- 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.)
- Active
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 123
- 230000004927 fusion Effects 0.000 title claims abstract description 48
- 238000000034 method Methods 0.000 title claims abstract description 42
- 238000012360 testing method Methods 0.000 claims abstract description 30
- 238000012549 training Methods 0.000 claims description 29
- 239000013598 vector Substances 0.000 claims description 22
- 238000013527 convolutional neural network Methods 0.000 claims description 16
- 238000012545 processing Methods 0.000 claims description 16
- 238000012706 support-vector machine Methods 0.000 claims description 15
- 238000013507 mapping Methods 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 10
- 238000007477 logistic regression Methods 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 10
- 238000004422 calculation algorithm Methods 0.000 claims description 9
- 238000010276 construction Methods 0.000 claims description 6
- 238000000354 decomposition reaction Methods 0.000 claims description 6
- 238000000605 extraction Methods 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 4
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 8
- 230000005540 biological transmission Effects 0.000 description 6
- 230000004913 activation Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000010801 machine learning Methods 0.000 description 4
- 238000005070 sampling Methods 0.000 description 4
- 230000000007 visual effect Effects 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 108010076504 Protein Sorting Signals Proteins 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000008094 contradictory effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Landscapes
- Test And Diagnosis Of Digital Computers (AREA)
Abstract
The invention discloses a line fault diagnosis method and device based on heterogeneous model fusion, wherein the method comprises the following steps: acquiring a line fault traveling wave signal of a test sample, and analyzing the fault traveling wave signal; respectively extracting time domain features, frequency domain features, time-frequency features and semantic features of the line fault traveling wave signals; constructing a single fault pre-diagnosis model based on the time domain features, the frequency domain features, the time-frequency features and the semantic features of the line fault traveling wave signals respectively, and performing preliminary diagnosis on the types of line faults of the test samples by using the fault pre-diagnosis model; and based on the D-S evidence theory, fusing each preliminary diagnosis result of the fault pre-diagnosis model, and constructing a fault diagnosis fusion model so as to carry out final diagnosis on the line fault type. The technical scheme of the invention can improve the accuracy of line fault classification under the condition of small sample size, and is beneficial to determining the line fault classification.
Description
Technical Field
The present invention relates to the field of line fault diagnosis technologies, and in particular, to a method and an apparatus for diagnosing a line fault based on heterogeneous model fusion, an electronic device, and a readable storage medium.
Background
With the rapid increase of power load demands, the scale of a power system is increasingly enlarged, the transmission distance and the voltage level are increased, the safe and stable operation of a power grid is extremely important, and a plurality of requirements for relay protection of the power system are increased. Because primary energy and load centers in China are unevenly distributed, a large-capacity long-distance power transmission task needs to be completed through a power transmission line. The transmission line is used as an important component from the power generation side to the power transmission and distribution side, and plays a vital role in the safe and stable operation of the power system. In recent years, transmission line accidents caused by weather disasters, external damage and line aging frequently occur, and the probability of line faults is higher than that of other components. When a line fails, fault current can be generated, the fault is rapidly and accurately identified through the fault current, and processing measures are taken, so that the power failure range is reduced, the normal operation of the system is recovered, and the necessary condition of system stability is ensured.
The traditional line fault classification method, such as a threshold method, a fuzzy logic reasoning method and the like, needs a great deal of expert experience and has the problem of insufficient classification precision. In recent years, with the development of artificial intelligence technology, some scholars have proposed a line fault classification technology based on a machine learning method, but often adopt a single type of feature to perform machine learning, and need to train with a large amount of analog data to achieve a sufficiently accurate effect, in practice, the line fault sample size often falls short of the required training data size, and the actual data is more complex and noisy than the analog data, so that the existing line fault classification method based on artificial intelligence has a problem of inaccurate prediction precision.
In view of this, there is a need for further improvements to the current technology.
Disclosure of Invention
In order to solve at least one of the above problems, a main object of the present invention is to provide a line fault diagnosis method, device, electronic apparatus and readable storage medium for heterogeneous model fusion.
In order to achieve the above object, the first technical scheme adopted by the present invention is as follows: the utility model provides a line fault diagnosis method based on heterogeneous model fusion, which comprises the following steps:
Acquiring a line fault traveling wave signal of a test sample, and analyzing the fault traveling wave signal;
respectively extracting time domain features, frequency domain features, time-frequency features and semantic features of the line fault traveling wave signals;
constructing a single fault pre-diagnosis model based on the time domain features, the frequency domain features, the time-frequency features and the semantic features of the line fault traveling wave signals respectively, and performing preliminary diagnosis on the types of line faults of the test samples by using the fault pre-diagnosis model;
And based on the primary diagnosis result of the D-S evidence theory fusion fault pre-diagnosis model, constructing a fault diagnosis fusion model to carry out final diagnosis on the line fault type.
The D-S evidence theory-based primary diagnosis result of the fault pre-diagnosis model is fused, and a calculation formula for constructing the fault diagnosis fusion model is as follows:
Where n=4, n is the number of feature fields, T j is the class of fault, m i(Tj) is the probability that the i-th single heterogeneous line fault pre-diagnosis model predicts that the test sample belongs to the class T j, C m represents the class of fault after feature field fusion, R (C m) represents the probability that the test sample belongs to the class C m after four feature fields fusion, and K represents the collision factor.
The method comprises the steps of constructing a single fault pre-diagnosis model based on time domain features, frequency domain features, time-frequency features and semantic features of the line fault traveling wave signals respectively, and outputting a category of line faults for preliminary diagnosis, and specifically comprises the following steps:
Inputting the extracted time domain features as feature vectors into a logistic regression model for training, and constructing a line fault pre-diagnosis model based on the time domain features and logistic regression;
Inputting the extracted frequency domain features as feature vectors into a support vector machine for training, and constructing a line fault pre-diagnosis model based on the frequency domain features and the support vector machine;
The extracted time-frequency characteristics are used as characteristic vectors to be input into a deep belief network for training, and a line fault pre-diagnosis model based on the time-frequency characteristics and the deep belief network is constructed; and
Inputting the extracted semantic features serving as feature vectors into a convolutional neural network for training, and constructing a line fault pre-diagnosis model based on the semantic features and the convolutional neural network;
And diagnosing the test sample by utilizing each fault pre-diagnosis model, and outputting the posterior probability of the corresponding category.
Wherein the time domain features, the frequency domain features, the time-frequency features and the semantic features of the line fault traveling wave signals are respectively extracted, the method specifically comprises the following steps:
Extracting time sequence characteristics of the line fault traveling wave signals, wherein the time sequence characteristics comprise maximum value, minimum value, extremely poor, variance, root mean square and kurtosis of the signals;
Performing fast Fourier transform on the fault traveling wave signals to obtain frequency domain characteristics;
carrying out wavelet packet decomposition on the fault traveling wave signals to obtain time-frequency characteristics; and
And processing the fault traveling wave signals by utilizing a symbiotic matrix algorithm, and extracting semantic features of the fault traveling wave signals.
The processing the fault traveling wave signal by utilizing the symbiotic matrix algorithm, and extracting the semantic features of the fault traveling wave signal specifically comprises the following steps:
Carrying out normalization processing on the fault traveling wave signals to enable the amplitude values of the fault traveling wave signals to be in the [ -1,1] interval;
Mapping the fault traveling wave signals after normalization processing, dividing [ -1,1] into intervals with preset quantity according to each interval of 0.01, and mapping all the amplitude values of the fault traveling wave signals to the corresponding intervals to obtain semantic features.
The step of obtaining the line fault traveling wave signal of the test sample and analyzing the fault traveling wave signal specifically comprises the following steps:
And acquiring a fault traveling wave and mat type file when the line fails, analyzing the fault traveling wave and mat type file according to the file format specification, and extracting a time sequence signal.
After analyzing the fault traveling wave and mat type file according to the file format specification and extracting the time sequence signal, the method further comprises the following steps:
the timing signals are stored in a. Txt type file and are marked for fault type.
In order to achieve the above object, the second technical scheme adopted by the present invention is as follows: provided is a line fault diagnosis device based on heterogeneous model fusion, comprising:
the acquisition module is used for acquiring a line fault traveling wave signal of the test sample and analyzing the fault traveling wave signal;
The extraction module is used for respectively extracting time domain features, frequency domain features, time-frequency features and semantic features of the line fault traveling wave signals;
The construction module is used for constructing a single fault pre-diagnosis model based on the time domain features, the frequency domain features, the time-frequency features and the semantic features of the line fault traveling wave signals respectively, and performing preliminary diagnosis on the types of line faults of the test samples by using the fault pre-diagnosis model;
And the diagnosis module is used for fusing the primary diagnosis result of the fault pre-diagnosis model based on the D-S evidence theory and constructing a fault diagnosis fusion model so as to carry out final diagnosis on the line fault type.
In order to achieve the above object, a third technical scheme adopted by the present invention is as follows: there is provided an electronic device including: the computer program comprises a memory, a processor and a computer program stored in the memory and capable of running on the processor, wherein the steps in the method are realized when the processor executes the computer program.
In order to achieve the above object, a fourth technical scheme adopted by the present invention is as follows: there is provided a readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the above method.
The technical scheme of the invention is that a line fault traveling wave signal of a test sample is firstly obtained, and the fault traveling wave signal is analyzed; and then respectively extracting time domain features, frequency domain features, time-frequency features and semantic features of the line fault traveling wave signals, respectively constructing a single fault pre-diagnosis model based on the time domain features, the frequency domain features, the time-frequency features and the semantic features of the line fault traveling wave signals, carrying out preliminary diagnosis on the types of line faults of the test samples by utilizing the fault pre-diagnosis model, finally fusing the preliminary diagnosis results of the fault pre-diagnosis model based on the D-S evidence theory, constructing a fault diagnosis fusion model to carry out final diagnosis on the types of the line faults, and carrying out fusion diagnosis on the faults and diagnosis models corresponding to a plurality of feature domains of the fault traveling wave signals by constructing output signals of each fault and diagnosis model according to the D-S evidence theory, so as to determine the final fault types of the test samples, thereby greatly improving the accuracy of the line fault classification under small sample size and being beneficial to determining the line fault types.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to the structures shown in these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for diagnosing line faults based on heterogeneous model fusion;
FIG. 2 is a block diagram of a deep belief network of the present invention;
FIG. 3 is a block diagram of a convolutional neural network of the present invention;
FIG. 4 is a block diagram of a circuit fault diagnosis device based on heterogeneous model fusion according to the present invention;
fig. 5 is a block diagram of the electronic device of the present invention.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are only some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
It should be noted that the description of "first," "second," etc. in this disclosure is for descriptive purposes only and is not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In addition, the technical solutions of the embodiments may be combined with each other, but it is necessary to base that the technical solutions can be realized by those skilled in the art, and when the technical solutions are contradictory or cannot be realized, the combination of the technical solutions should be considered to be absent and not within the scope of protection claimed in the present invention.
The invention provides a line fault diagnosis method based on heterogeneous model fusion, which is different from the line fault classification technology based on a machine learning method in the prior art, wherein single type features are adopted to perform machine learning, a large amount of simulation data are required to be adopted to train so as to achieve a sufficiently accurate effect, and in practice, the line fault sample size often cannot meet the requirement of the required training data size, so that the line fault classification is inaccurate. For specific steps of the line fault diagnosis method based on heterogeneous model fusion, please refer to the following embodiments.
Referring to fig. 1, fig. 1 is a flowchart of a line fault diagnosis method based on heterogeneous model fusion according to the present invention. In the embodiment of the invention, the line fault diagnosis method based on heterogeneous model fusion is mainly applied to diagnosing the line fault of the power line system. Specifically, the method comprises the following steps:
S110, obtaining a line fault traveling wave signal of a test sample, and analyzing the fault traveling wave signal. When a circuit in the power system fails, a corresponding line fault traveling wave signal is generated, and the line fault traveling wave signal is stored through a specific file (such as a mat type file and the like) so as to facilitate the subsequent determination of the line fault type.
Specifically, the obtaining the line fault traveling wave signal of the test sample, and analyzing the fault traveling wave signal specifically includes: and acquiring a fault traveling wave and mat type file when the line fails, analyzing the fault traveling wave and mat type file according to the file format specification, and extracting a time sequence signal.
More specifically, after analyzing the fault traveling wave and the mat type file according to the file format specification and extracting the time sequence signal, the method further comprises the following steps:
the timing signals are stored in a. Txt type file and are marked for fault type.
S120, respectively extracting time domain features, frequency domain features, time-frequency features and semantic features of the line fault traveling wave signals. In the step, four characteristic fields are selected for determining the line fault types, and the line fault types can be determined on the basis of not increasing the cost and considering the accuracy through the four characteristic fields. It will be appreciated that the four feature fields contain substantially all of the feature types that a signal can extract.
Specifically, the extracting the time domain feature, the frequency domain feature, the time-frequency feature and the semantic feature of the line fault traveling wave signal respectively specifically includes:
Extracting time sequence characteristics of the line fault traveling wave signals, including maximum value, minimum value, extremely poor, variance, root mean square and kurtosis of the signals. Specifically, the fault traveling wave time sequence signal is subjected to time sequence feature extraction, and the maximum value x max, the minimum value x min, the range R, the variance Var, the root mean square RMS and the kurtosis Kur of the extracted signals are extracted. Wherein,
xmax＝max(xi)，
xmin＝min(xi)，
R＝xmax-xmin,
Where σ is the standard deviation of the timing signal X, X i is the timing signal sequence point, and N is the sequence point.
And performing fast Fourier transform on the fault traveling wave signals to obtain frequency domain characteristics. Specifically, frequency domain feature extraction is performed on the fault traveling wave time sequence signal x= { X 1,x2,……,xn }, fourier transform is performed on the fault traveling wave time sequence signal, and the operation formula is as follows:
wherein x (k) represents a frequency value, and x n is a time sequence signal sequence point; n represents the sequence index of the time domain sampling points, and k represents the index of the time domain values; n represents the number of sampling points performed. The fourier transform is a simplification of the DFT, and if the complexity of the DFT is N 2 operations, the number of operations of the fourier transform is Nlg (N). Correspondence relation before and after conversion: assuming that the peak value of the original signal is A, the modulus value of each point (except the first point DC component) of the Fourier transform is N/2 times of A, and the frequency domain characteristic of the fault traveling wave is x fft＝{xfft1,xfft2,......,xfftn.
And carrying out wavelet packet decomposition on the fault traveling wave signals to obtain time-frequency characteristics. Specifically, the time-frequency characteristic extraction is performed on the fault traveling wave time sequence signal x= { X 1,x2,......,xn }, the wavelet packet decomposition is performed on the fault traveling wave time sequence signal, daubechies wavelet is adopted as a wavelet function base, and 8 coefficients obtained by performing three-layer wavelet packet decomposition are expressed as X w＝{xw1,xw2,......,xw8 }.
And processing the fault traveling wave signals by utilizing a symbiotic matrix algorithm, and extracting semantic features of the fault traveling wave signals. Specifically, semantic feature extraction is performed on the fault traveling wave time sequence signal X= { X 1,x2,......,xn }, and the fault traveling wave time sequence signal is processed by adopting a co-occurrence matrix algorithm to obtain semantic features of the fault traveling wave.
Further, the processing the fault traveling wave signal by using the co-occurrence matrix algorithm, and extracting the semantic features of the fault traveling wave signal specifically includes:
Carrying out normalization processing on the fault traveling wave signals to enable the amplitude values of the fault traveling wave signals to be in the [ -1,1] interval;
Mapping the fault traveling wave signals after normalization processing, dividing [ -1,1] into intervals with preset quantity according to each interval of 0.01, and mapping all the amplitude values of the fault traveling wave signals to the corresponding intervals to obtain semantic features.
Specifically, the fault traveling wave time sequence signal is normalized to the [ -1,1] interval, and the normalization formula is as follows:
Where x represents the original amplitude of the signal, |x| max represents the maximum value of the absolute value of the fault traveling wave timing signal, and x norm represents the timing signal after normalization. The feature sequence obtained after normalization of the original sequence is x norm＝{xnorm1,xnorm2,......,xnormn.
Mapping the normalized time sequence signal, dividing [ -1,1] into 200 intervals according to each interval of 0.01, mapping all amplitude values of the signal to the values of the corresponding intervals, for example, mapping the values of the intervals [ -1.00, -0.99] to be 1, mapping the values of the intervals [0.99,1] to be 200, and obtaining a characteristic sequence of x m＝{xm1,xm2,......,xmn }, wherein the characteristic sequence is obtained after mapping.
And constructing a symbiotic matrix of adjacent points by adopting the mapped sequence, wherein the size of the symbiotic matrix is 200 x 200, and rows and columns respectively represent the interval where the current point of the sequence is positioned and the interval where the next point is positioned, and the value of the symbiotic matrix is the frequency of occurrence. The specific method comprises the following steps: if the values of the points after the mapping of x normi and x normi+1 are x mi and x mi+1, the sum of all the points of the symbiotic matrix is n-1, that is, the n sampling points share n-1 logarithm of the adjacent points, as can be seen from the above description, the values of the rows x mi and the columns x mi+1 of the symbiotic matrix are added by 1.
S130, constructing a single fault pre-diagnosis model based on the time domain features, the frequency domain features, the time-frequency features and the semantic features of the line fault traveling wave signals respectively, and performing preliminary diagnosis on the line fault type of the test sample by using the fault pre-diagnosis model.
Specifically, the constructing a single fault pre-diagnosis model based on the time domain feature, the frequency domain feature, the time-frequency feature and the semantic feature of the line fault traveling wave signal respectively, and outputting a category of the line fault for preliminary diagnosis specifically includes:
And inputting the extracted time domain features serving as feature vectors into a logistic regression model for training, and constructing a line fault pre-diagnosis model based on the time domain features and the logistic regression.
And inputting the extracted frequency domain features serving as feature vectors into a support vector machine for training, and constructing a line fault pre-diagnosis model based on the frequency domain features and the support vector machine. The principle of the support vector machine is to find a segmentation hyperplane which can correctly classify data and has the largest distance, so that the performance of the support vector machine is often closely related to the selection of kernel functions. The support vector machine is a binary classifier, and adopts a one-to-one method to construct a multi-classifier, and the specific method is as follows: a support vector machine is designed between any two classes of samples, so k (k-1)/2 support vector machines are required for k classes of samples.
And inputting the extracted time-frequency characteristics as characteristic vectors into a deep belief network for training, and constructing a line fault pre-diagnosis model based on the time-frequency characteristics and the deep belief network. Specifically, the deep belief network model is based on a limited boltzmann machine RBM hierarchy, and each RBM hierarchy is composed of an input layer (visual layer) v and an output layer (hidden layer) h. Referring to fig. 2, fig. 2 is a block diagram of the deep belief network of the present invention. The deep belief network construction steps include:
And constructing a display layer unit v 1＝{v11,v12,…,v1n and a hidden layer unit h 1＝{h11,h12,…,h1n/2 of the RBM of the first-layer limited Boltzmann machine, wherein n and n/2 respectively represent the number of the display layer and the hidden layer unit, and the size of n is the size of the input feature vector.
And taking the hidden layer unit h 1 of the RBM layer of the first-layer limited Boltzmann machine as the display layer unit v 2 of the RBM layer of the second-layer limited Boltzmann machine, wherein the number of the hidden layer units of the RBM layer of the second-layer limited Boltzmann machine is 1/2 of that of the display layer units.
And constructing a third RBM layer of the restricted Boltzmann machine in the same way, and completing the construction of a three-layer RBM network model.
And adding a BP neural network at the top of the RBM layer of the third-layer limited Boltzmann machine for outputting classification results, wherein the number of output layer units is the number of fault types.
The limited boltzmann machine RBM can be regarded as a model of energy, which can be expressed as:
Wherein v i represents an input value of the display layer unit, h j represents an output value of the hidden layer unit, w ij represents a connection weight between the display layer unit v i and the hidden layer h j unit, a i represents a bias of the display layer unit v i, b j represents a bias of the hidden layer unit h j, m and n respectively represent node numbers of the display layer and the hidden layer unit, and θ= (a, b, w) form a model parameter of the restricted boltzmann machine RBM.
Defining the joint distribution probability of (v, h) according to an energy model as follows:
For m display layer units and n hidden layer units contained in a layer of limited boltzmann machine RBM, the conditional probability given a hidden layer unit and a display layer unit is:
the conditional probability of the hidden layer unit is:
according to the above formula, the activation probability of the hidden layer unit can be obtained as follows:
the activation probability of the display layer unit is:
Where σ () represents the activation function.
The deep belief network training process is as follows: for the first RBM layer of the deep belief network, the input is the frequency domain feature vector extracted by the fault traveling wave, and the output is the feature transformed by the RBM after pre-training; in the subsequent RBM layers, the input of the current RBM layer is the output of the previous RBM layer, and the output of the current RBM is the input of the next RBM layer; and for each RBM layer, performing pre-training by adopting a layer-by-layer unsupervised greedy learning algorithm, sampling and reconstructing the visual layer based on the activation probabilities of the display layer and the hidden layer, and adjusting the parameters of the RBM by utilizing the state of the visual layer and the error of the state of the reconstructed visual layer, so that the training of the RBM layer is stopped when the maximum iteration times or the reconstruction error meeting the requirements are reached. And respectively taking all the trained RBM parameters as initialization parameters of each layer of neural network, and then adopting the BP neural network to finely adjust the parameters so as to finally obtain the optimal parameters of the network.
And inputting the extracted semantic features serving as feature vectors into a convolutional neural network for training, and constructing a line fault pre-diagnosis model based on the semantic features and the convolutional neural network. Specifically, the convolutional neural network model includes a convolutional layer Conv, a pooling layer MaxPool, a ReLu layer, a fully connected layer FC, and a softmax layer. In the embodiment, 3 layers of convolution layers are adopted, each layer of convolution layer is followed by a layer Relu of nonlinear activation layer and a layer of pooling layer, the back of each layer of convolution layer is followed by a full-connection layer, the full-connection layer is followed by a softmax classification layer, and the output dimension of the softmax classification layer is the number of fault categories. Referring to fig. 3, fig. 3 is a block diagram of a convolutional neural network according to the present invention.
And training the convolutional neural network model through training set semantic feature data. The method specifically comprises the following steps: inputting data into the convolutional neural network model to obtain an output result, and judging whether the error between the output result and a real label value is in a threshold range or not; if yes, training is completed; and if not, adopting a cross entropy loss function and an Adam optimizer to adjust the convolutional neural network model until the error between the output result and the target reference value is within the threshold range.
And diagnosing the test sample by utilizing each fault pre-diagnosis model, and outputting the posterior probability of the corresponding category.
S140, based on the primary diagnosis result of the D-S evidence theory fusion fault pre-diagnosis model, constructing a fault diagnosis fusion model to carry out final diagnosis on the line fault type.
Under the condition of small sample size, the scheme can greatly improve the accuracy of line fault classification, and is favorable for determining line fault types.
The D-S evidence theory-based primary diagnosis result of the fault pre-diagnosis model is fused, and a calculation formula for constructing the fault diagnosis fusion model is as follows:
Where n=4, n is the number of feature fields, T j is the class of fault, m i(Tj) is the probability that the i-th single heterogeneous line fault pre-diagnosis model predicts that the test sample belongs to the class T j, C m represents the class of fault after feature field fusion, R (C m) represents the probability that the test sample belongs to the class C m after four feature fields fusion, and K represents the collision factor.
Referring to fig. 4, fig. 4 is a block diagram of a circuit fault diagnosis device based on heterogeneous model fusion according to the present invention. In an embodiment of the present invention, the line fault diagnosis apparatus based on heterogeneous model fusion includes:
An obtaining module 210, configured to obtain a line fault traveling wave signal of a test sample, and analyze the fault traveling wave signal;
the extracting module 220 is configured to extract time domain features, frequency domain features, time-frequency features and semantic features of the line fault traveling wave signal respectively;
The construction module 230 is configured to construct a single fault pre-diagnosis model based on the time domain feature, the frequency domain feature, the time-frequency feature and the semantic feature of the line fault traveling wave signal, and perform preliminary diagnosis on the line fault type of the test sample by using the fault pre-diagnosis model;
The diagnosis module 240 is configured to construct a fault diagnosis fusion model based on the primary diagnosis result of the D-S evidence theory fusion fault pre-diagnosis model, so as to perform final diagnosis on the line fault type.
Wherein, the construction module 230 is specifically configured to:
Inputting the extracted time domain features as feature vectors into a logistic regression model for training, and constructing a line fault pre-diagnosis model based on the time domain features and logistic regression;
Inputting the extracted frequency domain features as feature vectors into a support vector machine for training, and constructing a line fault pre-diagnosis model based on the frequency domain features and the support vector machine;
The extracted time-frequency characteristics are used as characteristic vectors to be input into a deep belief network for training, and a line fault pre-diagnosis model based on the time-frequency characteristics and the deep belief network is constructed; and
Inputting the extracted semantic features serving as feature vectors into a convolutional neural network for training, and constructing a line fault pre-diagnosis model based on the semantic features and the convolutional neural network;
And diagnosing the test sample by utilizing each fault pre-diagnosis model, and outputting the posterior probability of the corresponding category.
The extracting module 220 is specifically configured to:
Extracting time sequence characteristics of the line fault traveling wave signals, wherein the time sequence characteristics comprise maximum value, minimum value, extremely poor, variance, root mean square and kurtosis of the signals;
Performing fast Fourier transform on the fault traveling wave signals to obtain frequency domain characteristics;
carrying out wavelet packet decomposition on the fault traveling wave signals to obtain time-frequency characteristics; and
And processing the fault traveling wave signals by utilizing a symbiotic matrix algorithm, and extracting semantic features of the fault traveling wave signals.
Wherein, the extracting module 220 is further configured to:
Carrying out normalization processing on the fault traveling wave signals to enable the amplitude values of the fault traveling wave signals to be in the [ -1,1] interval;
Mapping the fault traveling wave signals after normalization processing, dividing [ -1,1] into intervals with preset quantity according to each interval of 0.01, and mapping all the amplitude values of the fault traveling wave signals to the corresponding intervals to obtain semantic features.
The obtaining module 210 is specifically configured to:
And acquiring a fault traveling wave and mat type file when the line fails, analyzing the fault traveling wave and mat type file according to the file format specification, and extracting a time sequence signal.
Wherein, the obtaining module 210 is further configured to:
the timing signals are stored in a. Txt type file and are marked for fault type.
Referring to fig. 5, fig. 5 is a block diagram of an electronic device according to the present invention. The electronic equipment can be used for realizing the line fault diagnosis method based on heterogeneous model fusion in the embodiment. As shown in fig. 5, the electronic device mainly includes: memory 301, processor 302, bus 303, and a computer program stored on memory 301 and executable on processor 302, memory 301 and processor 302 being connected by bus 303. When the processor 302 executes the computer program, the line fault diagnosis method based on heterogeneous model fusion in the foregoing embodiment is implemented. Wherein the number of processors may be one or more.
The memory 301 may be a high-speed random access memory (RAM, random Access Memory) memory or a non-volatile memory (non-volatile memory), such as a disk memory. The memory 301 is used for storing executable program code, and the processor 302 is coupled to the memory 301.
Further, the embodiment of the present invention further provides a readable storage medium, which may be provided in the electronic device in each of the foregoing embodiments, and the readable storage medium may be a memory in the foregoing embodiment shown in fig. 5.
The readable storage medium stores a computer program which, when executed by a processor, implements the line fault diagnosis method based on heterogeneous model fusion in the foregoing embodiment. Further, the computer-readable medium may be any medium capable of storing a program code, such as a usb (universal serial bus), a removable hard disk, a Read-Only Memory (ROM), a RAM, a magnetic disk, or an optical disk.
In the several embodiments provided by the present invention, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of modules is merely a logical function division, and there may be additional divisions of actual implementation, e.g., multiple modules or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or modules, which may be in electrical, mechanical, or other forms.
The modules illustrated as separate components may or may not be physically separate, and components shown as modules may or may not be physical modules, i.e., may be located in one place, or may be distributed over a plurality of network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist alone physically, or two or more modules may be integrated into one module. The integrated modules may be implemented in hardware or in software functional modules.
The integrated modules, if implemented in the form of software functional modules and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied essentially or in part or all of the technical solution or in part in the form of a software product stored in a readable storage medium, including instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of the embodiments of the present invention. And the aforementioned readable storage medium includes: a usb disk, a removable hard disk, a ROM, a RAM, a magnetic disk, or an optical disk, etc.
It should be noted that, for the sake of simplicity of description, the foregoing method embodiments are all expressed as a series of combinations of actions, but it should be understood by those skilled in the art that the present invention is not limited by the order of actions described, as some steps may be performed in other order or simultaneously in accordance with the present invention. Further, those skilled in the art will appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily all required for the present invention.
In the foregoing embodiments, the descriptions of the embodiments are emphasized, and for parts of one embodiment that are not described in detail, reference may be made to the related descriptions of other embodiments.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, and all the structural equivalents of the invention described in the specification and drawings are included in the scope of the invention, or the invention may be directly/indirectly applied to other related technical fields.
Claims (8)
1. The circuit fault diagnosis method based on heterogeneous model fusion is characterized by comprising the following steps of:
acquiring a line fault traveling wave signal and analyzing the fault traveling wave signal;
respectively extracting time domain features, frequency domain features, time-frequency features and semantic features of the line fault traveling wave signals;
Constructing a single fault pre-diagnosis model based on a time domain feature, a frequency domain feature, a time-frequency feature and a semantic feature of the line fault traveling wave signal respectively, and performing preliminary diagnosis information on the line fault type of the test sample by using the fault pre-diagnosis model, wherein the method specifically comprises the following steps: inputting the extracted time domain features as feature vectors into a logistic regression model for training, and constructing a line fault pre-diagnosis model based on the time domain features and logistic regression; inputting the extracted frequency domain features as feature vectors into a support vector machine for training, and constructing a line fault pre-diagnosis model based on the frequency domain features and the support vector machine; the extracted time-frequency characteristics are used as characteristic vectors to be input into a deep belief network for training, and a line fault pre-diagnosis model based on the time-frequency characteristics and the deep belief network is constructed; inputting the extracted semantic features as feature vectors into a convolutional neural network for training, and constructing a line fault pre-diagnosis model based on the semantic features and the convolutional neural network; diagnosing the test sample by utilizing each fault pre-diagnosis model, and outputting posterior probability of the corresponding category;
based on the output information of the D-S evidence theory fusion fault pre-diagnosis model, constructing a fault diagnosis fusion model to carry out final diagnosis on the line fault type, wherein the calculation formula of the fault diagnosis fusion model is as follows:
Wherein n=4, n is the number of feature domains, T j is the class of fault, m i(Tj) is the probability that the i-th single heterogeneous line fault pre-diagnosis model predicts that the test sample belongs to the class T j, C m represents the class of fault after feature domain fusion, and R (C m) represents the probability that the test sample belongs to the class C m after four feature domain fusion.
2. The method for diagnosing a line fault based on heterogeneous model fusion as claimed in claim 1, wherein the extracting time domain features, frequency domain features, time-frequency features and semantic features of the line fault traveling wave signals respectively comprises:
Extracting time sequence characteristics of the line fault traveling wave signals, wherein the time sequence characteristics comprise maximum value, minimum value, extremely poor, variance, root mean square and kurtosis of the signals;
Performing fast Fourier transform on the fault traveling wave signals to obtain frequency domain characteristics;
carrying out wavelet packet decomposition on the fault traveling wave signals to obtain time-frequency characteristics; and
And processing the fault traveling wave signals by utilizing a symbiotic matrix algorithm, and extracting semantic features of the fault traveling wave signals.
3. The line fault diagnosis method based on heterogeneous model fusion according to claim 2, wherein the processing the fault traveling wave signal by using a co-occurrence matrix algorithm, the extracting the semantic features of the fault traveling wave signal specifically comprises:
normalizing the fault traveling wave signal to a [ -1,1] interval;
Mapping the fault traveling wave signals after normalization processing, dividing [ -1,1] into intervals with preset quantity according to each interval of 0.01, and mapping all the amplitude values of the fault traveling wave signals to the corresponding intervals to obtain semantic features.
4. The method for diagnosing a line fault based on heterogeneous model fusion according to claim 1, wherein the steps of obtaining a line fault traveling wave signal and analyzing the fault traveling wave signal specifically include:
And acquiring a fault traveling wave and mat type file when the line fails, analyzing the fault traveling wave and mat type file according to the file format specification, and extracting a time sequence signal.
5. The line fault diagnosis method based on heterogeneous model fusion as claimed in claim 4, wherein after parsing the fault traveling wave mat type file according to the file format specification and extracting the timing signal, further comprising:
the timing signals are stored in a. Txt type file and are marked for fault type.
6. The utility model provides a circuit fault diagnosis device based on heterogeneous model fuses which characterized in that, circuit fault diagnosis device based on heterogeneous model fuses includes:
the acquisition module is used for acquiring a line fault traveling wave signal and analyzing the fault traveling wave signal;
The extraction module is used for respectively extracting time domain features, frequency domain features, time-frequency features and semantic features of the line fault traveling wave signals;
The construction module is used for constructing a single fault pre-diagnosis model based on the time domain feature, the frequency domain feature, the time-frequency feature and the semantic feature of the line fault traveling wave signal respectively, and carrying out preliminary diagnosis information on the line fault type of the test sample by utilizing the fault pre-diagnosis model, and specifically comprises the following steps: inputting the extracted time domain features as feature vectors into a logistic regression model for training, and constructing a line fault pre-diagnosis model based on the time domain features and logistic regression; inputting the extracted frequency domain features as feature vectors into a support vector machine for training, and constructing a line fault pre-diagnosis model based on the frequency domain features and the support vector machine; the extracted time-frequency characteristics are used as characteristic vectors to be input into a deep belief network for training, and a line fault pre-diagnosis model based on the time-frequency characteristics and the deep belief network is constructed; inputting the extracted semantic features as feature vectors into a convolutional neural network for training, and constructing a line fault pre-diagnosis model based on the semantic features and the convolutional neural network; diagnosing the test sample by utilizing each fault pre-diagnosis model, and outputting posterior probability of the corresponding category;
The diagnosis module is used for fusing output information of the fault pre-diagnosis model based on the D-S evidence theory and constructing a fault diagnosis fusion model so as to carry out final diagnosis on the line fault type, wherein the calculation formula of the fault diagnosis fusion model is as follows:
Wherein n=4, n is the number of feature domains, T j is the class of fault, m i(Tj) is the probability that the i-th single heterogeneous line fault pre-diagnosis model predicts that the test sample belongs to the class T j, C m represents the class of fault after feature domain fusion, and R (C m) represents the probability that the test sample belongs to the class C m after four feature domain fusion.
7. An electronic device, comprising: memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the method according to any one of claims 1 to 5 when the computer program is executed.
8. A readable storage medium having stored thereon a computer program, which, when executed by a processor, implements the steps of the method of any of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210020481.7A CN114358085B (en) | 2022-01-10 | 2022-01-10 | Line fault diagnosis method and device based on heterogeneous model fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210020481.7A CN114358085B (en) | 2022-01-10 | 2022-01-10 | Line fault diagnosis method and device based on heterogeneous model fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114358085A CN114358085A (en) | 2022-04-15 |
| CN114358085B true CN114358085B (en) | 2024-08-13 |
Family
ID=81107546
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210020481.7A Active CN114358085B (en) | 2022-01-10 | 2022-01-10 | Line fault diagnosis method and device based on heterogeneous model fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114358085B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117092445A (en) * | 2023-10-19 | 2023-11-21 | 盛隆电气集团有限公司 | Fault detection method and system of power distribution system based on big data |
| CN117668751B (en) * | 2023-11-30 | 2024-04-26 | 广东一业建设股份有限公司 | High-low voltage power system fault diagnosis method and device |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110617960A (en) * | 2019-10-12 | 2019-12-27 | 华北电力大学 | Wind turbine generator gearbox fault diagnosis method and system |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107894564B (en) * | 2017-11-09 | 2020-02-18 | 合肥工业大学 | Analog circuit fault diagnosis method based on cross wavelet characteristics |
| CN110766137A (en) * | 2019-10-18 | 2020-02-07 | 武汉大学 | Power electronic circuit fault diagnosis method based on longicorn whisker optimized deep confidence network algorithm |
| CN111370027B (en) * | 2020-03-02 | 2023-04-07 | 乐鑫信息科技（上海）股份有限公司 | Off-line embedded abnormal sound detection system and method |
-
2022
- 2022-01-10 CN CN202210020481.7A patent/CN114358085B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110617960A (en) * | 2019-10-12 | 2019-12-27 | 华北电力大学 | Wind turbine generator gearbox fault diagnosis method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114358085A (en) | 2022-04-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11549985B2 (en) | Power electronic circuit fault diagnosis method based on extremely randomized trees and stacked sparse auto-encoder algorithm | |
| Zhu et al. | Estimation of bearing remaining useful life based on multiscale convolutional neural network | |
| CN114358085B (en) | Line fault diagnosis method and device based on heterogeneous model fusion | |
| CN109974782B (en) | Equipment fault early warning method and system based on big data sensitive characteristic optimization selection | |
| EP4053751A1 (en) | Method and apparatus for training cross-modal retrieval model, device and storage medium | |
| CN113702895B (en) | Online quantitative evaluation method for error state of voltage transformer | |
| CN108520301A (en) | A kind of circuit intermittent fault diagnostic method based on depth confidence network | |
| CN108919059A (en) | A kind of electric network failure diagnosis method, apparatus, equipment and readable storage medium storing program for executing | |
| CN113295413B (en) | Traction motor bearing fault diagnosis method based on indirect signals | |
| CN115269870A (en) | Method for realizing classification and early warning of data link faults in data based on knowledge graph | |
| CN115758208A (en) | Traction converter fault diagnosis method and device, computer equipment and storage medium | |
| CN112990259A (en) | Early fault diagnosis method of rotary mechanical bearing based on improved transfer learning | |
| Zheng et al. | An unsupervised transfer learning method based on SOCNN and FBNN and its application on bearing fault diagnosis | |
| CN117219124B (en) | Switch cabinet voiceprint fault detection method based on deep neural network | |
| CN116910573B (en) | Training method and device for abnormality diagnosis model, electronic equipment and storage medium | |
| CN110110426A (en) | A kind of Switching Power Supply filter capacitor abatement detecting method | |
| CN114860945B (en) | High-quality noise detection method and device based on rule information | |
| CN116359738A (en) | Method, device, equipment and storage medium for monitoring health state of battery | |
| CN114397521A (en) | Fault diagnosis method and system for electronic equipment | |
| CN113821401A (en) | WT-GA-GRU model-based cloud server fault diagnosis method | |
| Chu et al. | A relaxed support vector data description algorithm based fault detection in distribution systems | |
| Jeske et al. | Mining and tracking massive text data: Classification, construction of tracking statistics, and inference under misclassification | |
| CN108763728B (en) | Mechanical fault diagnosis method using parallel deep neural network hierarchical feature extraction | |
| Mengjiao et al. | Wind turbine bearing fault diagnosis method based on multi-domain feature extraction | |
| CN110826690A (en) | Equipment state identification method and system and computer readable storage medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |