CN108919059A - A kind of electric network failure diagnosis method, apparatus, equipment and readable storage medium storing program for executing - Google Patents
A kind of electric network failure diagnosis method, apparatus, equipment and readable storage medium storing program for executing Download PDFInfo
- Publication number
- CN108919059A CN108919059A CN201810966929.8A CN201810966929A CN108919059A CN 108919059 A CN108919059 A CN 108919059A CN 201810966929 A CN201810966929 A CN 201810966929A CN 108919059 A CN108919059 A CN 108919059A
- Authority
- CN
- China
- Prior art keywords
- data
- neural network
- network model
- fault
- power grid
- 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 88
- 238000000034 method Methods 0.000 title claims abstract description 65
- 238000003062 neural network model Methods 0.000 claims abstract description 86
- 238000012549 training Methods 0.000 claims abstract description 17
- 230000006870 function Effects 0.000 claims description 57
- 210000002569 neuron Anatomy 0.000 claims description 35
- 238000004590 computer program Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 11
- 230000000694 effects Effects 0.000 abstract description 4
- 239000013598 vector Substances 0.000 description 18
- 230000008569 process Effects 0.000 description 8
- 230000008901 benefit Effects 0.000 description 5
- 230000002860 competitive effect Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000013507 mapping Methods 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 238000005259 measurement Methods 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 230000005284 excitation Effects 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000011524 similarity measure Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/081—Locating faults in cables, transmission lines, or networks according to type of conductors
- G01R31/086—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution networks, i.e. with interconnected conductors
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
-
- 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
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The invention discloses a kind of electric network failure diagnosis methods, on the one hand, obtaining sample data corresponding with power grid, and after according to predefined rule determining target data, noise signal is added for target data to form fault data, that is the method for diagnosing faults considers the influence of noise data, and then improves electric network failure diagnosis accuracy；On the other hand, after determining fault data, SOM neural network model is constructed according to remaining non-faulting data in fault data and sample data, and be trained to SOM neural network model；Finally electric network fault is diagnosed according to the SOM neural network model after training because SOM neural network model compared with the prior art in traditional BP neural network model, convergence rate is very fast, so further improving electric network failure diagnosis accuracy.In addition, the invention also discloses a kind of electric network failure diagnosis device, equipment and readable storage medium storing program for executing, effect is as above.
Description
Technical Field
The invention relates to the field of power system application, in particular to a power grid fault diagnosis method, a device, equipment and a readable storage medium.
Background
With the gradual construction of the smart power grid, the problem of fault diagnosis of the power grid is also concerned more and more, and operation and maintenance personnel can know the current operation condition of the power grid according to information displayed in the system, so as to analyze, locate and process faults, so that the identification of root faults is very important for the operation and maintenance work of the power grid. In recent years, fault diagnosis algorithms based on artificial intelligence are widely applied to power grid fault diagnosis, and aim to accurately analyze and find faults occurring in the power grid operation process.
At present, power grid fault information is mainly diagnosed by constructing a BP neural network model, but because the network data volume of a power grid system is large, the noise data is more, the fault types of the power grid faults are more, the convergence rate of the traditional BP neural network model is low, the influence of the noise data is not considered, and the accuracy of diagnosing the power grid faults is low.
Therefore, how to overcome the problem of low diagnosis accuracy when diagnosing the power grid fault is a problem to be solved urgently by technical personnel in the field.
Disclosure of Invention
The embodiment of the application provides a power grid fault diagnosis method, a device, equipment and a readable storage medium, which aim to solve the problem of low diagnosis accuracy when a power grid fault is diagnosed in the prior art.
In order to solve the technical problem, the invention provides a power grid fault diagnosis method, which comprises the following steps:
acquiring sample data corresponding to the power grid, and determining target data according to a predefined rule;
adding a noise signal to the target data to form fault data;
constructing an SOM neural network model according to the fault data and the non-fault data, and training the SOM neural network model;
diagnosing the power grid fault according to the trained SOM neural network model;
wherein the sample data is composed of the failure data and the non-failure data.
Preferably, after the obtaining of the sample data corresponding to the power grid, the method further includes:
and carrying out standardization processing on the sample data.
Preferably, the adding of the noise signal to the target data specifically includes:
and adding a Gaussian noise signal and a zero noise signal to the target data.
Preferably, after the building the SOM neural network model according to the fault data and the non-fault data, the method further includes:
a winning neuron is calculated and a neighborhood function for the winning neuron is determined.
Preferably, the determining a neighborhood function of the winning neuron specifically includes:
when the neighborhood range is equal to the threshold value, the neighborhood function is a coincidence function of a Cauchy function and a Gaussian function;
when the neighborhood range is greater than the threshold, the neighborhood function is the Cauchy function;
when the neighborhood range is less than the threshold, the neighborhood function is the Gaussian function.
Preferably, the diagnosing the grid fault according to the trained SOM neural network model specifically includes:
and diagnosing the power grid fault according to the trained similarity function in the SOM neural network model.
In order to solve the above technical problem, the present invention further provides a device corresponding to the power grid fault diagnosis method, including:
the acquisition module is used for acquiring sample data corresponding to the power grid and determining target data according to a predefined rule;
an adding module, configured to add a noise signal to the target data to form fault data;
the building module is used for building an SOM neural network model according to the fault data and the non-fault data and training the SOM neural network model;
the diagnosis module is used for diagnosing the power grid fault according to the trained SOM neural network model;
wherein the sample data is composed of the failure data and the non-failure data.
Preferably, the method further comprises the following steps:
and the data processing module is used for carrying out standardized processing on the sample data.
In order to solve the technical problem, the present invention further provides a device corresponding to the power grid fault diagnosis method, including:
a memory for storing a computer program;
a processor for executing the computer program to implement the steps of any one of the above-mentioned grid fault diagnosis methods.
In order to solve the technical problem, the present invention further provides a device computer readable storage medium corresponding to the power grid fault diagnosis method, where the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement any one of the steps of the power grid fault diagnosis method.
Compared with the prior art, the power grid fault diagnosis method provided by the invention has the advantages that on one hand, after the sample data corresponding to the power grid is obtained and the target data is determined according to the predefined rule, the noise signal is added to the target data to form the fault data, namely the influence of the noise data is considered by the fault diagnosis method, so that the power grid fault diagnosis accuracy is improved; on the other hand, after the fault data are determined, an SOM neural network model is constructed according to the fault data and the residual non-fault data in the sample data, and the SOM neural network model is trained; and finally, diagnosing the power grid fault according to the trained SOM neural network model, wherein the SOM neural network model has higher convergence speed compared with the traditional BP neural network model in the prior art, so that the power grid fault diagnosis accuracy is further improved. In addition, the invention also provides a power grid fault diagnosis device, equipment and a readable storage medium, and the effect is as above.
Drawings
Fig. 1 is a flowchart of a power grid fault diagnosis method according to an embodiment of the present invention;
fig. 2 is a schematic diagram illustrating a power grid fault diagnosis apparatus according to an embodiment of the present invention;
fig. 3 is a schematic diagram illustrating a power grid fault diagnosis device according to an embodiment of the present invention;
fig. 4 is a graph illustrating a relationship between the grid fault diagnosis accuracy and the alarm cascade provided by the embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The core of the invention is to provide a power grid fault diagnosis method, a device, equipment and a readable storage medium, which can solve the problem of low diagnosis accuracy when diagnosing power grid faults in the prior art.
In order that those skilled in the art will better understand the concept of the present invention, the following detailed description of the invention is provided in conjunction with the accompanying drawings and the detailed description of the invention.
Fig. 1 is a flowchart of a power grid fault diagnosis method according to an embodiment of the present invention, and as shown in fig. 1, the diagnosis method includes the following steps:
s101: and acquiring sample data corresponding to the power grid, and determining target data according to a predefined rule.
S102: noise signals are added to the target data to form fault data.
Specifically, sample data corresponding to the power grid is selected, and then target data is determined from the selected sample data according to a percentage (predefined rule), wherein the target data is data needing to be processed. When the target data is determined, a noise signal is added to the target data to form fault data, namely, normal target data is changed into abnormal fault data to consider the influence of noise disturbance on a diagnosis result. Because the difference between the actually measured data and the data in the standard experiment is very large, the situation that the actually measured data is interfered by noise is very common, and the phenomena that the data has abnormal values and missing values are more.
In order to make the prediction accuracy of the grid fault higher, as a preferred embodiment, after obtaining sample data corresponding to the grid, the method further includes:
and carrying out standardization processing on the sample data.
S103: and constructing an SOM neural network model according to the fault data and the non-fault data, and training the SOM neural network model.
S104: and diagnosing the power grid fault according to the trained SOM neural network model.
Wherein the sample data is composed of failure data and non-failure data.
Firstly, target data is selected from sample data and processed into fault data, residual data except the target data in the sample data become non-fault data, then an SOM neural network model is constructed according to the fault data and the non-fault data, and the constructed SOM neural network model is trained. And in the later stage, the power grid fault can be diagnosed directly according to the trained SOM neural network model without processing the acquired power grid related data.
The SOM neural network model (self-organizing map neural network model) is a competitive neural network characterized by preserving the high-dimensional features of the input vectors and mapping them onto the low-dimensional neural network, thereby forming a topologically ordered graph.
In the embodiment of the application, in order to train the noise reduction capability of the SOM model, that is, in consideration of the influence of the noise disturbance data on the diagnostic result, a certain noise disturbance needs to be added in the process of training the SOM model. As a preferred embodiment, adding the noise signal to the target data is specifically: adding Gaussian noise signals to target dataSign and zero noise signal. Specifically, a random noise layer is added in front of an input layer of the SOM neural network model, sample data is processed, and part of data (target data) is selected to be added with Gaussian noise and set with 0 noise randomly, so that a competitive layer is forced to learn more robust features in the data. The structure of the SOM neural network model in the embodiment of the application is composed of an input layer, a random noise layer and a competition layer. The input layer randomly acquires a group of sample data vectors (subjected to data normalization processing) as the input of the SOM neural network model. The random noise layer carries out random noise adding mapping q on input sample data X to obtain partially destroyed target dataThe mapping process is shown as follows:
where D represents the vector space.
In practical application, for each sample data vector, the noise mapping layer randomly selects target data with fixed number of dimensionalities and adds Gaussian noise signals or zero setting signals. For example, target data with a ratio of 30% may be selected from sample data and a noise signal may be added thereto, and in the case where the data size is large and the network model is complicated, the ratio of the noise-added target data may be appropriately reduced. If the experimental data is less, in order to better mine the potential features of the data and improve the robustness of the model, the proportion of the noise-added target data should be improved.
The working modes of the random noise layer are mainly divided into two types: one is to set zero noise signal, set the selected target data of sample data vector to 0, simulate the data missing condition; the other is a gaussian noise signal, which is added to the selected target data to change its probability distribution. The specific formula of the gaussian distribution is as follows:
where σ denotes the sample data variance, μ denotes the sample data mean, and X denotes an attribute of the sample data X. The noisy map may be expressed as:
where d is a linear random number.
The input layer of the SOM neural network model is responsible for receiving input data, and the number of neurons in the input layer is the dimension number of one sample datum. The number of neurons in the competition layer is usually chosen to be approximately the square of the dimension of the input data and arranged in a matrix to form a plane. Each contention layer neuron maintains a weight vector WjWhere j represents the sequence number of the neuron, the dimension of the weight vector is the same as the sample data dimension. The input layer is fully connected to each of the competition layer neurons, while the competition layer neurons are not connected to each other (note: there are also connections between competition layer neurons in some variants of the SOM neural network model), and the mechanism of operation of the SOM neural network model is the competition learning rule. Firstly, sample data vectors are normalized in a European space, and weight vectors are initialized randomly in the European space. Then, randomly selecting an input vector and all weight vectors to carry out similarity measurement, and selecting the neuron with the highest similarity as a winning neuron, wherein the formula is as follows:
where d represents a similarity measure, X represents an input vector, j*A winning neuron is represented by a number of neurons,a weight vector. Final activation of winning neurons andneurons in the neighborhood, and neurons far from the winning neuron are suppressed, so that the competing layer nodes retain the topological features of the input vector. Through the repeated training, the neuron of the final competition layer tends to the clustering center, and a standard is provided for data classification.
In practical application, in order to improve the accuracy of power grid fault diagnosis, in the process of classifying data, the used similarity measurement of the SOM neural network model is to find the euclidean distance, and the missing and abnormal of some dimension attributes in the data sample vector can greatly affect the euclidean distance, so that the data is misclassified, namely fault misdiagnosis. In order to solve the problem that the robustness of a fault diagnosis task is low due to noise interference, a noise reduction idea is introduced into an SOM neural network model. In addition, the competition layer of the SOM neural network model performs similarity measurement by calculating euclidean distances to select the winning neuron that most closely approximates the input data. For noisy data, calculating the euclidean distance results in a large error. Therefore, in the embodiment of the present application, a fuzzy euclidean distance calculation method is designed to reduce noise influence, so that the calculation process of the competition layer is more robust, and the calculation formula is as follows:
wherein,representing the mean of X elements of the sample data, W representing the weight of the sample data, WiWeight, x, representing the ith sample dataiThe ith sample data is represented.
The process of training the SOM neural network model comprises the following steps: in the first step, the number of training times is set to determine whether to end the current training. And secondly, inputting power grid data to detect faults. And thirdly, considering the problem of data noise, and adding a random noise layer to the SOM neural network model to simulate noise and interference in a real environment. Fourth, a winning neuron is calculated according to the competitive neural network principle and a winning neighborhood is defined. And fifthly, gradually reducing the learning efficiency and adjusting the weighting coefficient of the data according to the iteration times. And sixthly, calculating the training times, if the required times are reached, ending, otherwise, returning to the first step.
On one hand, after sample data corresponding to the power grid is obtained and target data are determined according to predefined rules, noise signals are added to the target data to form fault data, namely the influence of the noise data is considered by the fault diagnosis method, and therefore the accuracy of power grid fault diagnosis is improved; on the other hand, after the fault data are determined, an SOM neural network model is constructed according to the fault data and the residual non-fault data in the sample data, and the SOM neural network model is trained; and finally, diagnosing the power grid fault according to the trained SOM neural network model, wherein the SOM neural network model has higher convergence speed compared with the traditional BP neural network model in the prior art, so that the power grid fault diagnosis accuracy is further improved.
On the basis of the above embodiment, as a preferred implementation, after constructing the SOM neural network model according to the fault data and the non-fault data, the method further includes:
a winning neuron is computed and a neighborhood function for the winning neuron is determined.
On the basis of the foregoing embodiment, as a preferred implementation, the determining a neighborhood function of the winning neuron specifically includes:
when the neighborhood range is equal to the threshold value, the neighborhood function is a coincidence function of the Cauchy function and the Gaussian function;
when the neighborhood range is larger than the threshold value, the neighborhood function is a Cauchy function;
when the neighborhood range is less than the threshold, the neighborhood function is a gaussian function.
The class boundary is an important parameter and represents the range of the influence of the winning neuron of the competition layer of the SOM neural network model. The neighborhood function is an important way that the SOM neural network model uses to fit class boundaries.
Generally, when the SOM neural network model is used, a gaussian function is used as a neighborhood function, and the euclidean distance between a neuron and a current winning neuron is calculated to determine the participation degree of a common neuron in a training process, so that the convergence rate of the SOM neural network model algorithm is higher than that of a general rectangular neighborhood function, but the excitation effect of the SOM neural network model algorithm in a far-neighborhood range is not obvious. The excitation effect of the Cauchy function on a near neighborhood is far less good than that of a Gaussian function, but the Cauchy function has better output capability on a far neighborhood range. Therefore, the embodiment of the application designs a composite neighborhood function, and integrates the advantages of the Gaussian function in a near domain and the advantages of the Cauchy function in a far-adjacent domain.
The Cauchy function and the Gaussian function are combined into a complex function by taking sigma (t) as a boundary, namely, the Gaussian function is selected in the vicinity of the complex function, and the Cauchy function is selected in the far vicinity of the complex function.
When sigma (t) < 3, a Gaussian function is used as a neighborhood function of the winning neuron, and the expression is as follows:
when sigma (t) > 3, the Cauchy function is used as the neighborhood function of the winning neuron, and the expression is as follows:
||rc-ri||2representing the euclidean distance between node c and node r, σ (t) representing the neighborhood range, which is a decreasing function with respect to t, and becomes progressively smaller as t increases.
In order to accurately locate the type of the grid fault, on the basis of the above embodiment, as a preferred implementation, diagnosing the grid fault according to the trained SOM neural network model specifically includes:
and diagnosing the power grid fault according to the similarity function in the trained SOM neural network model.
Specifically, because the fault type of the grid fault may be various, that is, there are multiple source faults, the multiple source fault diagnosis means that one alarm data may contain multiple root fault information in order to detect all relevant root fault information. A similarity function between the alarm sample data and the cluster center needs to be established to measure the similarity between one input sample and the cluster. Therefore, in the diagnosis of the multiple source faults, the main task is to select the root source alarm with high similarity and eliminate the root source alarm with low similarity, and the method can be specifically based on the following formula:
s(W,X)＝d(W,X)/∑d(W,X)
where Σ d (W, X) represents the distance between all cluster centers and the input data. In practical applications, k cluster centers with the smallest distance can be selected ifn represents the number of cluster centers, and k source faults are represented.
The above embodiment of the power grid fault diagnosis method is described in detail, and based on the power grid fault diagnosis method described in the above embodiment, the embodiment of the present invention further provides a power grid fault diagnosis apparatus corresponding to the method. Since the embodiment of the apparatus portion and the embodiment of the method portion correspond to each other, the embodiment of the apparatus portion is described with reference to the embodiment of the method portion, and is not described again here.
Fig. 2 is a schematic composition diagram of a power grid fault diagnosis apparatus according to an embodiment of the present invention, and as shown in fig. 2, the diagnosis apparatus includes an obtaining module 201, an adding module 202, a constructing module 203, and a diagnosis module 204.
The acquisition module 201 is configured to acquire sample data corresponding to a power grid and determine target data according to a predefined rule;
an adding module 202, configured to add a noise signal to the target data to form fault data;
the building module 203 is used for building an SOM neural network model according to the fault data and the non-fault data and training the SOM neural network model;
the diagnosis module 204 is used for diagnosing the power grid fault according to the trained SOM neural network model;
wherein the sample data is composed of failure data and non-failure data.
On one hand, the power grid fault diagnosis device provided by the invention has the advantages that after sample data corresponding to a power grid is obtained and target data is determined according to a predefined rule, a noise signal is added to the target data to form fault data, namely the influence of the noise data is considered by the fault diagnosis method, so that the power grid fault diagnosis accuracy is improved; on the other hand, after the fault data are determined, an SOM neural network model is constructed according to the fault data and the residual non-fault data in the sample data, and the SOM neural network model is trained; and finally, diagnosing the power grid fault according to the trained SOM neural network model, wherein the SOM neural network model has higher convergence speed compared with the traditional BP neural network model in the prior art, so that the power grid fault diagnosis accuracy is further improved.
On the basis of the above embodiment, as a preferred implementation, the method further includes:
and the data processing module is used for carrying out standardized processing on the sample data.
The above embodiment of the power grid fault diagnosis method is described in detail, and based on the power grid fault diagnosis method described in the above embodiment, the embodiment of the present invention further provides a power grid fault diagnosis device corresponding to the method. Since the embodiment of the device part and the embodiment of the method part correspond to each other, the embodiment of the device part is described with reference to the embodiment of the method part, and is not described again here.
Fig. 3 is a schematic diagram illustrating a power grid fault diagnosis device according to an embodiment of the present invention, and as shown in fig. 3, the diagnosis device includes a memory 301 and a processor 302.
A memory 301 for storing a computer program;
a processor 302 for executing a computer program to implement the steps of the grid fault diagnosis method provided by any of the above embodiments.
On one hand, the power grid fault diagnosis equipment provided by the invention has the advantages that after sample data corresponding to a power grid is obtained and target data is determined according to a predefined rule, a noise signal is added to the target data to form fault data, namely the influence of the noise data is considered by the fault diagnosis method, so that the power grid fault diagnosis accuracy is improved; on the other hand, after the fault data are determined, an SOM neural network model is constructed according to the fault data and the residual non-fault data in the sample data, and the SOM neural network model is trained; and finally, diagnosing the power grid fault according to the trained SOM neural network model, wherein the SOM neural network model has higher convergence speed compared with the traditional BP neural network model in the prior art, so that the power grid fault diagnosis accuracy is further improved.
The foregoing describes in detail an embodiment of a power grid fault diagnosis method, and based on the power grid fault diagnosis method described in the foregoing embodiment, an embodiment of the present invention further provides a computer-readable storage medium corresponding to the method. Since the embodiment of the computer-readable storage medium portion and the embodiment of the method portion correspond to each other, please refer to the embodiment of the method portion for describing the embodiment of the computer-readable storage medium portion, which is not described herein again.
A computer-readable storage medium, on which a computer program is stored, the computer program being executed by a processor to implement the steps of the grid fault diagnosis method provided by any one of the above embodiments.
In the computer-readable storage medium provided by the invention, the processor can read a program stored in the readable storage medium, that is, the power grid fault diagnosis method provided by any one of the embodiments can be implemented, on one hand, after sample data corresponding to a power grid is obtained and target data is determined according to a predefined rule, a noise signal is added to the target data to form fault data, that is, the fault diagnosis method takes the influence of the noise data into consideration, so that the power grid fault diagnosis accuracy is improved; on the other hand, after the fault data are determined, an SOM neural network model is constructed according to the fault data and the residual non-fault data in the sample data, and the SOM neural network model is trained; and finally, diagnosing the power grid fault according to the trained SOM neural network model, wherein the SOM neural network model has higher convergence speed compared with the traditional BP neural network model in the prior art, so that the power grid fault diagnosis accuracy is further improved.
In order to make those skilled in the art better understand the present solution, the present solution is described below in terms of a practical application scenario, and a simulation comparison experiment is performed on the SOM neural network model by using simulated power alarm data. The electric power alarm data is in a numerical value vector form, the data does not contain specific information of the root fault, but the data with the same root fault has certain similarity in a vector space. In the training stage, the data of the training set is input into the SOM neural network model for clustering, after clustering is finished, each activated competitive layer neuron represents a clustering center, and the root fault type of the training data of the cluster is the fault type represented by the clustering center. In the testing stage, if the single source fault diagnosis task is performed, the alarm data to be tested is input into the SOM neural network model to obtain a winning neuron, and in practical application, the type of the root source fault represented by the winning neuron can be considered as the root source fault of the alarm data to be tested; if the task is a multi-source fault diagnosis task, the fault type needs to be positioned according to the similarity function.
In the embodiment of the application, a simulation data set is designed according to the characteristics of power grid data, and ten network faults such as antenna abnormity, load exceeding limit, sensor abnormity, electromagnetic interference and the like are simulated. The data set mode is structured data, and each piece of data comprises multidimensional attributes such as power consumption, sensor statistical data and flow, and is normalized into a numerical vector. Thus, the data itself does not represent a network failure, but data with similar failures meets the clustering criteria. In the analysis of the electric power alarm data, it is assumed that the accuracy of fault diagnosis has a certain relationship with the cascade number of alarms, i.e., the alarm data can trigger multiple levels of alarms. A comparison experiment of the accuracy of root fault diagnosis under different cascade alarm conditions is designed. Fig. 4 is a graph illustrating a relationship between accuracy of power grid fault diagnosis and alarm cascade, according to an embodiment of the present invention, as shown in fig. 4, a solid line in fig. 4 represents a curve illustrating a relationship between accuracy of power grid fault diagnosis and alarm cascade, which is diagnosed by using a trained SOM neural network model after considering a noise signal; the dotted line represents a curve of the relation between the power grid fault accuracy diagnosed by using the trained SOM neural network model and the alarm cascade when the noise signal is not considered, and the fault diagnosis accuracy tends to increase along with the increase of cascade. The reason is that when the cascade of root cause alarm triggers is small, its interference resistance against noise is weak. And because the SOM neural network model introduces a noise reduction mechanism, a higher precision ratio can be kept under the condition of less cascade numbers.
The method, the device, the equipment and the readable storage medium for diagnosing the power grid fault provided by the invention are described in detail above. The principles and embodiments of the present invention have been described herein using several examples, the above description of which is only intended to facilitate the understanding of the method and its core concepts of the present invention; meanwhile, for those skilled in the art, based on the idea of the present invention, there may be variations in the specific embodiments and applications, and in summary, the present disclosure should not be construed as a limitation of the present invention, and those skilled in the art should include modifications, equivalent substitutions, improvements and the like without inventive labor.
It is further noted that, in the present specification, relational terms such as first and second, and the like are used solely to distinguish one from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Furthermore, the term "comprises/comprising" and the like, such that a unit, device or system comprising a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such unit, device or system.
Claims (10)
1. A power grid fault diagnosis method is characterized by comprising the following steps:
acquiring sample data corresponding to the power grid, and determining target data according to a predefined rule;
adding a noise signal to the target data to form fault data;
constructing an SOM neural network model according to the fault data and the non-fault data, and training the SOM neural network model;
diagnosing the power grid fault according to the trained SOM neural network model;
wherein the sample data is composed of the failure data and the non-failure data.
2. The grid fault diagnosis method according to claim 1, further comprising, after the obtaining of the sample data corresponding to the grid:
and carrying out standardization processing on the sample data.
3. The grid fault diagnosis method according to claim 1, wherein the adding of the noise signal to the target data is specifically:
and adding a Gaussian noise signal and a zero noise signal to the target data.
4. The grid fault diagnosis method according to claim 1, further comprising, after the building of the SOM neural network model from the fault data and non-fault data:
a winning neuron is calculated and a neighborhood function for the winning neuron is determined.
5. The grid fault diagnosis method according to claim 4, wherein the determining the neighborhood function of the winning neuron specifically comprises:
when the neighborhood range is equal to the threshold value, the neighborhood function is a coincidence function of a Cauchy function and a Gaussian function;
when the neighborhood range is greater than the threshold, the neighborhood function is the Cauchy function;
when the neighborhood range is less than the threshold, the neighborhood function is the Gaussian function.
6. The power grid fault diagnosis method according to claim 1, wherein the diagnosing of the power grid fault according to the trained SOM neural network model specifically comprises:
and diagnosing the power grid fault according to the trained similarity function in the SOM neural network model.
7. A grid fault diagnosis apparatus, comprising:
the acquisition module is used for acquiring sample data corresponding to the power grid and determining target data according to a predefined rule;
an adding module, configured to add a noise signal to the target data to form fault data;
the building module is used for building an SOM neural network model according to the fault data and the non-fault data and training the SOM neural network model;
the diagnosis module is used for diagnosing the power grid fault according to the trained SOM neural network model;
wherein the sample data is composed of the failure data and the non-failure data.
8. The grid fault diagnosis device according to claim 7, further comprising:
and the data processing module is used for carrying out standardized processing on the sample data.
9. A grid fault diagnosis apparatus, characterized by comprising:
a memory for storing a computer program;
a processor for executing the computer program to implement the steps of the grid fault diagnosis method according to any one of claims 1 to 6.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program which is executed by a processor to implement the steps of the grid fault diagnosis method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810966929.8A CN108919059A (en) | 2018-08-23 | 2018-08-23 | A kind of electric network failure diagnosis method, apparatus, equipment and readable storage medium storing program for executing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810966929.8A CN108919059A (en) | 2018-08-23 | 2018-08-23 | A kind of electric network failure diagnosis method, apparatus, equipment and readable storage medium storing program for executing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108919059A true CN108919059A (en) | 2018-11-30 |
Family
ID=64405582
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810966929.8A Pending CN108919059A (en) | 2018-08-23 | 2018-08-23 | A kind of electric network failure diagnosis method, apparatus, equipment and readable storage medium storing program for executing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108919059A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110244188A (en) * | 2019-07-10 | 2019-09-17 | 南方电网科学研究院有限责任公司 | Power grid fault diagnosis method and device and power distribution switch monitoring terminal |
| CN110716843A (en) * | 2019-09-09 | 2020-01-21 | 深圳壹账通智能科技有限公司 | System fault analysis processing method and device, storage medium and electronic equipment |
| CN111199252A (en) * | 2019-12-30 | 2020-05-26 | 广东电网有限责任公司 | Fault diagnosis method for intelligent operation and maintenance system of power communication network |
| WO2020125166A1 (en) * | 2018-12-20 | 2020-06-25 | 北京科东电力控制系统有限责任公司 | Power grid anticipated fault set prediction method and apparatus, and electronic device and storage medium |
| CN111461392A (en) * | 2020-01-23 | 2020-07-28 | 华中科技大学 | Power failure prediction method and system based on graph neural network |
| CN111580506A (en) * | 2020-06-03 | 2020-08-25 | 南京理工大学 | Industrial process fault diagnosis method based on information fusion |
| CN112884068A (en) * | 2021-03-15 | 2021-06-01 | 大庆油田有限责任公司 | Fault diagnosis method and device, electronic equipment and storage medium |
| CN113642624A (en) * | 2021-08-05 | 2021-11-12 | 厦门大学 | Intelligent diagnosis method and device for mobile communication network fault |
| CN114740288A (en) * | 2022-03-10 | 2022-07-12 | 云南电网有限责任公司电力科学研究院 | Lightning arrester fault diagnosis method and device, electronic equipment and storage medium |
| CN115684835A (en) * | 2022-10-28 | 2023-02-03 | 贵州电网有限责任公司信息中心 | Power distribution network fault diagnosis method, device, equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104330255A (en) * | 2014-11-04 | 2015-02-04 | 徐州隆安光电科技有限公司 | Gear fault diagnosing method based on multi-sensor information fusion |
| CN104808107A (en) * | 2015-04-16 | 2015-07-29 | 国家电网公司 | XLPE cable partial discharge defect type identification method |
| CN106443310A (en) * | 2016-11-22 | 2017-02-22 | 国网四川省电力公司广安供电公司 | Transformer fault detection method based on SOM (Self Organizing Map) neural network |
| CN107367647A (en) * | 2017-06-22 | 2017-11-21 | 上海理工大学 | The detection of mains by harmonics source and localization method based on EEMD SOM |
| CN107957551A (en) * | 2017-12-12 | 2018-04-24 | 南京信息工程大学 | Stacking noise reduction own coding Method of Motor Fault Diagnosis based on vibration and current signal |
| CN108152651A (en) * | 2017-12-27 | 2018-06-12 | 重庆水利电力职业技术学院 | Transmission line malfunction integrated recognition method based on GMAPM and SOM-LVQ-ANN |
-
2018
- 2018-08-23 CN CN201810966929.8A patent/CN108919059A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104330255A (en) * | 2014-11-04 | 2015-02-04 | 徐州隆安光电科技有限公司 | Gear fault diagnosing method based on multi-sensor information fusion |
| CN104808107A (en) * | 2015-04-16 | 2015-07-29 | 国家电网公司 | XLPE cable partial discharge defect type identification method |
| CN106443310A (en) * | 2016-11-22 | 2017-02-22 | 国网四川省电力公司广安供电公司 | Transformer fault detection method based on SOM (Self Organizing Map) neural network |
| CN107367647A (en) * | 2017-06-22 | 2017-11-21 | 上海理工大学 | The detection of mains by harmonics source and localization method based on EEMD SOM |
| CN107957551A (en) * | 2017-12-12 | 2018-04-24 | 南京信息工程大学 | Stacking noise reduction own coding Method of Motor Fault Diagnosis based on vibration and current signal |
| CN108152651A (en) * | 2017-12-27 | 2018-06-12 | 重庆水利电力职业技术学院 | Transmission line malfunction integrated recognition method based on GMAPM and SOM-LVQ-ANN |
Non-Patent Citations (2)
| Title |
|---|
| 赵伟 等: "q-高斯的SOM 神经网络在雷达抗干扰效能评估中的应用", 《哈尔滨工程大学学报》 * |
| 郑建宏 等: "电力线通信系统中基于OFDM/OQAM的时频结合消噪算法", 《计算机应用》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020125166A1 (en) * | 2018-12-20 | 2020-06-25 | 北京科东电力控制系统有限责任公司 | Power grid anticipated fault set prediction method and apparatus, and electronic device and storage medium |
| CN110244188A (en) * | 2019-07-10 | 2019-09-17 | 南方电网科学研究院有限责任公司 | Power grid fault diagnosis method and device and power distribution switch monitoring terminal |
| CN110716843A (en) * | 2019-09-09 | 2020-01-21 | 深圳壹账通智能科技有限公司 | System fault analysis processing method and device, storage medium and electronic equipment |
| CN110716843B (en) * | 2019-09-09 | 2022-11-22 | 深圳壹账通智能科技有限公司 | System fault analysis processing method and device, storage medium and electronic equipment |
| CN111199252A (en) * | 2019-12-30 | 2020-05-26 | 广东电网有限责任公司 | Fault diagnosis method for intelligent operation and maintenance system of power communication network |
| CN111461392A (en) * | 2020-01-23 | 2020-07-28 | 华中科技大学 | Power failure prediction method and system based on graph neural network |
| CN111461392B (en) * | 2020-01-23 | 2022-06-17 | 华中科技大学 | Power failure prediction method and system based on graph neural network |
| CN111580506A (en) * | 2020-06-03 | 2020-08-25 | 南京理工大学 | Industrial process fault diagnosis method based on information fusion |
| CN112884068A (en) * | 2021-03-15 | 2021-06-01 | 大庆油田有限责任公司 | Fault diagnosis method and device, electronic equipment and storage medium |
| CN113642624A (en) * | 2021-08-05 | 2021-11-12 | 厦门大学 | Intelligent diagnosis method and device for mobile communication network fault |
| CN114740288A (en) * | 2022-03-10 | 2022-07-12 | 云南电网有限责任公司电力科学研究院 | Lightning arrester fault diagnosis method and device, electronic equipment and storage medium |
| CN115684835A (en) * | 2022-10-28 | 2023-02-03 | 贵州电网有限责任公司信息中心 | Power distribution network fault diagnosis method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108919059A (en) | A kind of electric network failure diagnosis method, apparatus, equipment and readable storage medium storing program for executing | |
| CN115018021B (en) | Machine room abnormity detection method and device based on graph structure and abnormity attention mechanism | |
| CN113255848B (en) | Water turbine cavitation sound signal identification method based on big data learning | |
| JP7126256B2 (en) | Abnormality diagnosis device, abnormality diagnosis method, and program | |
| CN116956197B (en) | Deep learning-based energy facility fault prediction method and device and electronic equipment | |
| CN106405384A (en) | Simulation circuit health state evaluation method | |
| CN114266289A (en) | Complex equipment health state assessment method | |
| CN113409166A (en) | XGboost model-based method and device for detecting abnormal electricity consumption behavior of user | |
| CN117269742A (en) | Method, device and medium for evaluating health state of circuit breaker in high-altitude environment | |
| CN116414653A (en) | Method and device for detecting host fault, electronic equipment and storage medium | |
| CN116595465A (en) | High-dimensional sparse data outlier detection method and system based on self-encoder and data enhancement | |
| CN114881157A (en) | Method, device and equipment for detecting working state of converter valve and storage medium | |
| CN115392782A (en) | Method and system for monitoring and diagnosing health state of process system of nuclear power plant | |
| CN118152567A (en) | Staff portrait anomaly detection method and terminal | |
| CN116627116B (en) | Process industry fault positioning method and system and electronic equipment | |
| CN116384223A (en) | Nuclear equipment reliability assessment method and system based on intelligent degradation state identification | |
| CN115375038A (en) | Aircraft engine failure mode identification and service life prediction method | |
| CN114676593A (en) | Abnormity detection method of textile equipment and related device | |
| CN113051809A (en) | Virtual health factor construction method based on improved restricted Boltzmann machine | |
| CN114137915A (en) | Fault diagnosis method for industrial equipment | |
| CN111597934A (en) | System and method for processing training data for statistical applications | |
| CN112541554B (en) | Multi-mode process monitoring method and system based on time constraint and nuclear sparse representation | |
| CN117609737B (en) | Method, system, equipment and medium for predicting health state of inertial navigation system | |
| CN114580101B (en) | Method and system for predicting residual service life of rotary machine | |
| CN118549823B (en) | Lithium battery electrical performance testing method and system |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181130 |
|
| RJ01 | Rejection of invention patent application after publication |