CN115902557A - Switch cabinet fault diagnosis processing method and device and nonvolatile storage medium - Google Patents
Switch cabinet fault diagnosis processing method and device and nonvolatile storage medium Download PDFInfo
- Publication number
- CN115902557A CN115902557A CN202211732267.0A CN202211732267A CN115902557A CN 115902557 A CN115902557 A CN 115902557A CN 202211732267 A CN202211732267 A CN 202211732267A CN 115902557 A CN115902557 A CN 115902557A
- Authority
- CN
- China
- Prior art keywords
- data
- partial discharge
- fault diagnosis
- fault
- initial
- 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
Images
Landscapes
- Testing Relating To Insulation (AREA)
Abstract
The invention discloses a switch cabinet fault diagnosis processing method and device and a nonvolatile storage medium. Wherein, the method comprises the following steps: acquiring initial partial discharge fault data of a target switch cabinet; preprocessing initial partial discharge fault data by adopting a local mean decomposition algorithm to obtain first partial discharge fault data; acquiring an initial fault diagnosis model constructed based on a multiplication gated cyclic neural network algorithm; and inputting the first partial fault data into the initial fault diagnosis model for training to obtain a target fault diagnosis model. The invention solves the technical problems of low fault diagnosis efficiency and poor diagnosis accuracy in the related technology.
Description
Technical Field
The invention relates to the technical field of insulation state evaluation of power equipment, in particular to a switch cabinet fault diagnosis processing method and device and a nonvolatile storage medium.
Background
The switch cabinet plays the important role of power transformation and transmission in daily operation, and many switch cabinets are placed in the open air in daily life, are exposed to the sun and rain for a long time, are influenced by various external environments, and are easy to have partial discharge faults. Moreover, in a complex environment where a power system is continuously maintained for a long time at high temperature and high pressure, or during the manufacturing, transportation and assembly of the switch cabinet, some safety hazards such as dust, conductive particles, metal tips, air gaps, etc. are inevitably generated, which may cause various types of partial discharge, thereby causing insulation failure or power system failure. Therefore, the potential correlation of identifying the partial discharge mode and the defect type is an important index in the insulation diagnosis of the gas-insulated metal-enclosed switchgear, the potential defects can be found in time, the generated defects can be solved, and the operation safety of the power system can be greatly guaranteed. However, the line pipe technology mainly adopts a manual detection mode for the fault diagnosis of the switch cabinet, and the detection efficiency is low. The identification mode based on the neural network algorithm has relatively poor accuracy of fault diagnosis of the switch cabinet and low fault diagnosis efficiency due to the reasons of algorithm performance or model practicability and the like.
In view of the above problems, no effective solution has been proposed.
Disclosure of Invention
The embodiment of the invention provides a switch cabinet fault diagnosis processing method and device and a nonvolatile storage medium, which are used for at least solving the technical problems of low fault diagnosis efficiency and poor diagnosis accuracy in the related technology.
According to an aspect of the embodiments of the present invention, there is provided a method for diagnosing and processing a fault of a switch cabinet, including: acquiring initial partial discharge fault data of a target switch cabinet; preprocessing the initial partial discharge fault data by adopting a local mean decomposition algorithm to obtain first partial discharge fault data; acquiring an initial fault diagnosis model constructed based on a multiplication gated cyclic neural network algorithm; and inputting the first partial fault data into the initial fault diagnosis model for training to obtain a target fault diagnosis model.
According to another aspect of the embodiments of the present invention, there is also provided a switch cabinet fault diagnosis processing apparatus, including: the first acquisition module is used for acquiring initial partial discharge fault data of the target switch cabinet; the processing module is used for preprocessing the initial partial discharge fault data by adopting a local mean decomposition algorithm to obtain first partial discharge fault data; the second acquisition module is used for acquiring an initial fault diagnosis model constructed based on a multiplicative gated recurrent neural network algorithm; and the training module is used for inputting the first partial fault data into the initial fault diagnosis model for training to obtain a target fault diagnosis model.
According to another aspect of the embodiments of the present invention, there is also provided a non-volatile storage medium, wherein the non-volatile storage medium stores a plurality of instructions, and the instructions are suitable for being loaded by a processor and executing any one of the switch cabinet fault diagnosis processing methods.
According to another aspect of the embodiments of the present invention, there is also provided an electronic device, including one or more processors and a memory, where the memory is used for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement any one of the switch cabinet fault diagnosis processing methods described above.
In the embodiment of the invention, initial partial discharge fault data of a target switch cabinet are obtained; preprocessing the initial partial discharge fault data by adopting a local mean decomposition algorithm to obtain first partial discharge fault data; acquiring an initial fault diagnosis model constructed based on a multiplication gated cyclic neural network algorithm; the first local fault data are input into the initial fault diagnosis model to be trained, a target fault diagnosis model is obtained, the purposes of improving the acquisition precision of the switch cabinet fault diagnosis model and accurately identifying the switch cabinet fault are achieved, the technical effects of optimizing the switch cabinet fault diagnosis model and improving the fault diagnosis efficiency and the fault diagnosis accuracy of the switch cabinet are achieved, and the technical problems of low fault diagnosis efficiency and poor diagnosis accuracy in the related technology are solved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the invention and do not constitute a limitation of the invention. In the drawings:
fig. 1 is a schematic diagram of a method for diagnosing and processing a fault of a switch cabinet according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of an alternative switchgear fault diagnosis processing method according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of a switchgear fault diagnosis processing system according to an embodiment of the present invention;
fig. 4 is a schematic diagram of a switch cabinet fault diagnosis processing device according to an embodiment of the invention.
Detailed Description
In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, 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.
It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in other sequences than those illustrated or described herein. Moreover, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
First, in order to facilitate understanding of the embodiments of the present invention, some terms or nouns referred to in the present invention will be explained as follows:
local Mean Decomposition (LMD), as a new adaptive time-frequency analysis method appearing in recent years, can decompose a complex multi-component am/fm signal into a sum of finite single-component am/fm signals according to the characteristics of the signal itself, and further obtain and combine instantaneous frequency and instantaneous amplitude to obtain complete time-frequency distribution of the original signal. Compared with the empirical mode decomposition method, the method has the advantages that the problems of end point effect, false components, over-envelope and under-envelope are improved.
Singular Value Decomposition (SVD), an important matrix decomposition method in linear algebra and matrix theory. The algorithm is widely applied to the fields of communication system MIMO, machine learning, image processing, data compression and noise reduction and the like.
In an electric power grid, a switch cabinet (switch cabinet) is an electrical device, an external line of the switch cabinet firstly enters a main control switch in the cabinet and then enters a branch control switch, and each branch circuit is arranged according to the requirement of the branch circuit. Such as meters, automatic controls, magnetic switches of motors, various alternating current contactors, and the like, some of which are also provided with high-voltage chambers and low-voltage chamber switch cabinets, and high-voltage buses, such as power plants, and some of which are also provided with low-frequency load shedding for keeping main equipment. The low-voltage power distribution cabinet is a general name of a motor control center in electrical equipment. The power distribution cabinet is suitable for occasions with relatively dispersed loads and small loops; the motor control center has centralized load and more circuit occasions. They distribute energy from one circuit of the upper level distribution equipment to the nearest load. This level of equipment should provide protection, monitoring and control of the load. If the power distribution cabinet breaks down in use, the use of the multi-stage equipment is directly influenced.
The switch cabinet plays the important role of power transformation and transmission in daily operation, and many switch cabinets are placed in the open air in daily life, are exposed to the sun and rain for a long time, are influenced by various external environments, and are easy to cause partial discharge faults. Moreover, in the complex environment of the power system with long-term high temperature and high pressure, or during the manufacturing, transportation and assembly of the switch cabinet, some safety hazards such as dust, conductive particles, metal tips, air gaps, etc. are inevitably generated, which may cause partial discharge in various forms, and thus cause insulation failure or power system failure. Therefore, the potential correlation of identifying the partial discharge mode and the defect type is an important index in the insulation diagnosis of the GIS, the potential defects can be found in time, the generated defects can be solved, and the operation safety of the power system is greatly guaranteed. However, the line pipe technology mainly adopts a manual detection mode for the fault diagnosis of the switch cabinet, and the detection efficiency is low.
In recent years, deep Neural Networks (DNNs) have been studied in many fields such as machine vision, speech recognition, and fault recognition and diagnosis. Meanwhile, the improvement of the calculation performance and the optimization of the algorithm also create good conditions for DNN, and the fault diagnosis method based on artificial intelligence becomes a popular research direction with higher classification accuracy rate of the fault diagnosis method on the operation state types of the power equipment. RNNS is unable to capture a wide range of sequence dependencies due to gradient disappearance and gradient explosion problems. To address this problem, LSTM arose by introducing a gating mechanism that captures a potentially large range of sequence dependencies by selectively retaining previous information, allowing important features to be detected from the input sequence, and retaining that feature information for a long period of time. The GRU is an improvement on the basis of the LSTM, the model simplifies the number of gating, can obtain better effect than the LSTM by using fewer parameters when constructing a larger network, and can save the calculation cost. Due to the reasons of algorithm performance or model practicability and the like, the switch cabinet fault detection and identification mode based on the neural network algorithm has relatively poor accuracy for switch cabinet fault diagnosis and low fault diagnosis efficiency.
In view of the above problems, it should be noted that the steps illustrated in the flowcharts of the figures may be implemented in a computer system such as a set of computer executable instructions, and that while a logical order is illustrated in the flowcharts, in some cases, the steps illustrated or described may be performed in an order different than here.
Fig. 1 is a flowchart of a fault diagnosis processing method for a switch cabinet according to an embodiment of the present invention, and as shown in fig. 1, the method includes the following steps:
step S102, obtaining initial partial discharge fault data of the target switch cabinet.
Optionally, the original data of the partial discharge fault of the switch cabinet is acquired as initial partial discharge fault data in a real-time acquisition mode of the partial discharge signal of the target switch cabinet. With the continuous use of the target switch cabinet equipment, the partial discharge fault data can be automatically updated and written in and gradually increased, and the accuracy of fault diagnosis and identification can be greatly improved by abundant sample data. And after the original data of the partial discharge fault of the switch cabinet, the integrity of the original data is required to be checked, if the data is missing, other data with the same type as the missing data in the original data is obtained, the median of the other data is taken to fill the missing data, and the filled original data of the partial discharge fault of the switch cabinet is used as the original partial discharge fault data.
And step S104, preprocessing the initial partial discharge fault data by adopting a local mean decomposition algorithm to obtain first partial discharge fault data.
It should be noted that the operating environment is complex, and various noises are doped in the collected partial discharge signal, so that the embodiment of the present invention performs noise reduction processing on the obtained initial partial discharge fault data by using a local mean decomposition algorithm, and performs subsequent model training by using the first partial discharge fault data after the noise reduction processing, thereby effectively avoiding interference of other factors such as noise on the fault diagnosis model construction, and facilitating obtaining of a more accurate and effective fault diagnosis model.
In an optional embodiment, the preprocessing the initial partial discharge fault data by using a local mean decomposition algorithm to obtain first partial discharge fault data includes: performing first noise reduction processing on the initial partial discharge fault data by adopting the local mean decomposition algorithm to obtain second partial discharge fault data; performing second noise reduction processing on the initial partial discharge fault data by adopting a singular value decomposition algorithm to obtain third partial discharge fault data; and performing feature extraction processing on the third partial discharge fault data to obtain the first partial discharge fault data.
Through the mode, firstly, the Local Mean Decomposition (LMD) algorithm is adopted to carry out primary noise reduction on initial partial discharge fault data, then, the Singular Value Decomposition (SVD) algorithm is adopted to carry out secondary noise reduction on residual noise in an LMD decomposition result, and the weighted energy contribution rate (PCTE) is used as a method for determining the singular value order, so that noise reduction processing is carried out on the PF component, and secondary filtering is realized.
In an optional embodiment, the performing, by using the local mean decomposition algorithm, a first noise reduction process on the initial partial discharge fault data to obtain a second partial discharge fault data includes: obtaining a plurality of product functions based on the local mean decomposition algorithm; taking the initial partial discharge fault data as an original signal, and determining correlation coefficients between the multiple product functions and the original signal respectively; determining a boundary position between a noise signal and an effective signal included in the original signal based on correlation coefficients between the plurality of product functions and the original signal, respectively; and eliminating the signals before the boundary position in the original signals, and taking the eliminated original signals as the second partial discharge fault data.
Through the above mode, firstly, LMD decomposition is used to obtain a series of Product Functions (PF) distributed from high frequency to low frequency; and determining the boundary position between the noise-containing signal and the effective signal by calculating the correlation coefficient between the original signal and each product function PF component, and eliminating the component before the boundary position in the initial partial discharge fault data to realize the initial noise reduction.
Optionally, for the LMD, the LMD is an adaptive time domain analysis algorithm, which decomposes a multi-component signal into a sum of a plurality of Product Functions (PFs) having actual physical quantity meanings, each PF component is a product of an envelope signal and a frequency modulation signal, wherein an instantaneous frequency of the PF component can be found from the frequency modulation signal, and an instantaneous amplitude of the PF component is the envelope signal, thereby obtaining a time domain distribution of an original signal. In the invention, for an original partial discharge fault signal x (t) of a switch cabinet, the LMD decomposition process is as follows:
determining all extreme points n of original signals x (t) in partial discharge of switch cabinet i (i =1,2.. Eta.), calculating a local mean function m of the i-th section i And a local envelope function a i Respectively is as follows:
processing the local mean function and the local envelope function by using a moving average method until the values of any adjacent points are unequal to obtain a local mean function m 11 (t) and a local envelope function a 11 (t)。
Will local mean function m 11 (t) separating from the partial discharge original signal x (t) to obtain a signal h 11 (t), namely:
h 11 (t)＝x(t)-m 11 (t)
by h 11 (t) divided by the local envelope function a 11 (t), for h 11 (t) demodulating to obtain a frequency modulated signal s 11 (t), namely:
multiplying all local envelope functions to obtain an envelope signal a 1 (t), namely:
envelope signal a 1 (t) with a pure FM signal s 1n (t) the multiplication results in the first PF component of the signal x (t), i.e.:
PF 1 (t)＝a 1 (t)s 1n (t)
from x (t) the component PF 1 (t) separating to obtain a new signal u 1 (t) using u 1 (t) repeating the above steps in place of the original signal and cycling K times until u K (t) monotonously.
The switch cabinet raw signal x (t) can thus be decomposed into K PF components and a monotonic function u K (t) sum, i.e.
And determining the dividing position between the noise-containing signal and the effective signal by calculating the correlation coefficient between the original signal and each PF component, and eliminating the components before the dividing component to realize primary noise reduction. Then, for residual noise in the LMD decomposition result, using an SVD method, and taking a weighted energy contribution to total energy (PCTE) as a determination method of a singular value order, performing noise reduction processing on the boundary PF component, and implementing secondary filtering. Assuming that a is the residual noise matrix in the LMD decomposition result, its singular value decomposition is:
wherein U is AA T Is a, V is A T A matrix formed by the eigenvectors of A, the value of the singular value matrix sigma being A T The square root of the a feature.
In an optional embodiment, the performing, by the third partial discharge fault data, a feature extraction process to obtain a feature extraction result as the first partial discharge fault data includes: normalizing the third partial discharge fault data to obtain fourth partial discharge fault data; and performing feature extraction processing on the fourth partial discharge fault data to obtain the first partial discharge fault data.
Through the above manner, after the third partial discharge fault data after noise reduction is obtained, the third partial discharge fault data is further normalized to accelerate the training speed, and the normalization formula is as follows:
wherein x is i For any one of the third partial discharge failure data, y i And obtaining a normalized result (corresponding to fourth partial discharge fault data) corresponding to any one data in the third partial discharge fault data.
Optionally, feature extraction is performed on the normalized data (i.e., fourth partial discharge fault data), and the correspondingly extracted features may be not limited to include: the characteristic value comprises positive half-cycle skewness of maximum discharge distribution
Negative half-cycle deflection for a maximum discharge profile>
The positive half-cycle projection of the maximum discharge profile->
The maximum discharge measure distribution negative half-cycle protrusion degree->
Maximum discharge distribution asymmetry Q m Maximum discharge quantity distribution correlation degree CC m The mean charge distribution is positive and the deflection is greater than or equal to>
Negative half-cycle skewness/mean discharge quantity distribution>
Mean discharge quantity distribution positive half-cycle projection->
Mean discharge quantity distribution negative half-cycle projection->
Mean discharge amount distribution asymmetry Q a Average discharge quantity distribution correlation degree CC a The degree of partial inclination of the positive half-cycle of the discharge frequency distribution>
Negative half-cycle deflection degree based on discharge frequency distribution>
The number of discharges is distributed over the positive half-cycle protrusion degree->
The number of discharges is distributed with a negative half-cycle projection degree->
Asymmetry Q of discharge frequency distribution n And the degree of correlation CC of the discharge frequency distribution n 18, the data structure is shown in the following table 1:
TABLE 1
Each group of characteristic values corresponds to one partial discharge fault type, and the partial discharge type of the switch cabinet may include, but is not limited to: internal (air gap) discharge, along-the-plane (surface) discharge, corona discharge, and levitation discharge.
And S106, acquiring an initial fault diagnosis model constructed based on a multiplicative gated cyclic neural network algorithm.
It should be noted that GRU (Gate recovery Unit) is a variant of LSTM network with good effect, which is simpler than LSTM network structure, and can also solve the long dependence problem in RNN network, and the effect is also good. A multiplicative gated recurrent neural network model (i.e., a multi-GRU network model) uses an element-by-element multiplication on the output from the last hidden state to determine what will be merged into the new hidden state at the current time step. The method has good applicability to fault diagnosis of the switch cabinet.
And S108, inputting the first local fault data into the initial fault diagnosis model for training to obtain a target fault diagnosis model.
In an optional embodiment, the inputting the first partial fault data into the initial fault diagnosis model for training to obtain a target fault diagnosis model includes: dividing the first partial release fault data into training set data and test set data; inputting the training set data into the initial fault diagnosis model for training to obtain a trained initial fault diagnosis model; inputting the test set data into the trained initial fault diagnosis model for testing to obtain a test result; and taking the trained initial fault diagnosis model as the target fault diagnosis model under the condition that the test result meets the preset test condition.
By the method, the initial fault diagnosis model is trained by adopting training set data in the first partial fault data, and the trained initial fault diagnosis model is output when the training reaches the preset iteration times; and testing the trained initial fault diagnosis model by using the test set data in the first local fault data, if the test result meets the preset test condition, indicating that the model meets the requirements, and taking the trained initial fault diagnosis model as a final target fault diagnosis model. The target fault diagnosis model is deployed in a field test environment, and data to be tested are input into the target fault diagnosis model, so that real-time monitoring of the switch cabinet can be achieved.
Optionally, a multi-GRU network model is established as an initial fault diagnosis model, the first local fault data includes 6000 groups of sample data, the data are randomly scrambled, the data are split according to 8. And transmitting the training set for training. The input-output structure of the multi-GRU is the same as a normal RNN. Let the input at time t be x t The hidden layer state at the time t-1 is h t-1 The hidden layer state contains the relevant information of the previous node; the output of the hidden node at the moment t is y t The implicit state passed to the next node is h t . Two gates, reset gate and refresh gate, are in the GRU model, the initial state, when t =0 is, the output vector h 0 And =0. The mathematical model of the multi-GRU is as follows:
z t ＝σ g (W z x t +U z h t-1 +b z )
r t ＝σ g (W r x t +U r h t-1 +b r )
wherein x is t As input vector, h t To pass on to the implicit state at the next moment,
as candidate hidden states, z t To refresh the door, r t For resetting the gate, W, U and b are parameter matrices, σ g Is a sigmoid function, phi h Is a function of tanh. The update gate z selects whether or not to use the new implicit state->
The implicit state is updated. The reset gate r decides whether to ignore the preceding implicit state. Here->
Mainly containing x of the current input t And (6) data. When r is close to 0, the implicit state forces the preceding implicit state to be ignored and reset with the current input. In targeted pairs>
Adding to the current implicit state corresponds to "memorizing the state at the current time". The range of the gate signal z is 0 to 1, and the closer the gate signal is to 1, the more data is "memorized", and the closer to 0, the more data is "forgotten". Sigma g The data can be changed to a value in the range of 0 to 1, phi h The data may be changed to a value ranging from-1 to 1.
The iteration number in the model training process is determined when an accuracy-iteration number curve tends to be stable, and 200 is taken here. After the model training is finished, testing and data are adopted to test the trained initial fault diagnosis model, and under the condition that the testing result meets the preset testing condition, the trained initial fault diagnosis model is used as the target fault diagnosis model, and partial discharge fault diagnosis of the switch cabinet can be carried out by inputting data to be diagnosed.
In an optional embodiment, after the inputting the first partial fault data into the initial fault diagnosis model for training to obtain a target fault diagnosis model, the method further includes: acquiring data to be diagnosed corresponding to a switch cabinet to be detected; and inputting the data to be diagnosed to the target fault diagnosis model for testing to obtain a fault diagnosis result.
Through the mode, the trained target fault diagnosis model is deployed to a field test environment, and when the switch cabinet is actually tested on site, the data to be diagnosed corresponding to the switch cabinet to be tested is input to the target fault diagnosis model, so that the field real-time fault monitoring of the switch cabinet to be tested can be realized.
Through the steps S102 to S108, local Mean Decomposition (LMD) algorithm is used for denoising partial discharge signals of the switch cabinet, true partial discharge data are obtained as far as possible, main characteristic values are extracted to serve as sample data and are transmitted into a multiplication gate control recurrent neural network (MultiGRU) model for training to obtain an ideal diagnosis result, the purpose of improving the acquisition precision of a fault diagnosis model of the switch cabinet and accurately identifying faults of the switch cabinet can be achieved, the technical effects of optimizing the fault diagnosis model of the switch cabinet, improving the fault diagnosis efficiency and the fault diagnosis accuracy of the switch cabinet are achieved, and the technical problems of low fault diagnosis efficiency and poor diagnosis accuracy in the related technology are solved.
Based on the foregoing embodiment and alternative embodiments, the present invention provides an alternative implementation manner, and fig. 2 is a flowchart of an alternative fault diagnosis processing method for a switch cabinet according to an embodiment of the present invention, as shown in fig. 2, the method includes:
s1, acquiring original partial discharge fault data of the switch cabinet as initial partial discharge fault data in a real-time partial discharge signal acquisition mode of a target switch cabinet. The method specifically comprises the following substeps:
step (1.1): with the continuous use of the target switch cabinet equipment, partial discharge fault data can be updated and written in by itself and is gradually increased, and the accuracy of fault diagnosis and identification can be greatly improved by abundant sample data;
step (1.2): after the original data of the partial discharge fault of the switch cabinet, the integrity of the original data is required to be checked, if the data is missing, other data with the same type as the missing data in the original data is obtained, the median of the other data is taken to fill the missing data, and the filled original data of the partial discharge fault of the switch cabinet is used as the original partial discharge fault data.
And S2, denoising the original data of the partial discharge fault of the switch cabinet by using a Local Mean Decomposition (LMD) algorithm, then carrying out normalization processing, and then extracting a main characteristic value as first partial discharge fault data. The method specifically comprises the following substeps:
step (2.1): firstly, using LMD decomposition to obtain a series of Product Functions (PF) distributed from high frequency to low frequency; and determining the boundary position between the noise-containing signal and the effective signal by calculating the correlation coefficient between the original signal and each product function PF component, and eliminating the component before the boundary position in the initial partial discharge fault data to realize the initial noise reduction. Then, aiming at the residual noise in the LMD decomposition result, performing secondary noise reduction on the residual noise in the LMD decomposition result by adopting a Singular Value Decomposition (SVD) algorithm, and performing noise reduction on the PF (boundary of distribution to total energy, PCTE) component by using a determination method of a singular value order to realize secondary filtering.
Step (2.2): and normalizing the denoised third partial discharge fault data to accelerate the training speed to obtain fourth partial discharge fault data.
Step (2.3): extracting characteristic values of fourth partial discharge fault data acquired after normalization, wherein the characteristic values comprise
Q m 、CC m 、/>
Q a 、CC a 、/>
Q n 、CC n 18 kinds of the Chinese medicinal materials.
And S3, building a multiplication gating recurrent neural network model (namely a multi-GRU network model) as an initial fault diagnosis model.
And S4, inputting the test set data in the first local fault data into a multi-GRU network model for training, and determining the iteration times.
S5, judging whether the model training meets the iteration requirement, and if so, outputting the trained multi-GRU network model; and inputting the data of the testing machine into the trained multi-GRU network model, and taking the obtained trained initial fault diagnosis model as a target fault diagnosis model under the condition that the output test result meets the preset test condition. Inputting the data to be diagnosed into a trained target fault diagnosis model, then outputting a diagnosis result, and if the iteration requirement is not met or the test result does not meet the preset test condition, continuing to train the model.
It should be noted that, because the operating environment of the switch cabinet is complex, various noises are doped in the collected partial discharge signals, and the signals are collected on site and then diagnosed on site, a large amount of labor cost is generated, and various potential safety hazards also exist. The method adopts the neural network in the field of artificial intelligence to solve the problems, firstly, the collected partial discharge signals are subjected to LMD algorithm denoising, then, after normalization processing and feature extraction, the partial discharge signals are transmitted into a multi-GRU network model provided by the invention for training, and finally, an ideal diagnosis result is obtained. The risk and the artificial misjudgment rate caused by field operation are avoided, a large amount of manpower is saved, the identification speed is increased, and the identification accuracy is improved.
According to an embodiment of the present invention, there is further provided a system embodiment for implementing the method for diagnosing and processing a fault of a switch cabinet, and fig. 3 is a schematic structural diagram of a system for diagnosing and processing a fault of a switch cabinet according to an embodiment of the present invention, and as shown in fig. 3, the system includes:
the remote server side is used for receiving the local discharge signals of the field switch cabinet through the remote server and realizing real-time updating of the local discharge signals;
the data processing module is connected with the remote server end and used for receiving the partial discharge signal data from the remote server and carrying out denoising, normalization, characteristic value extraction and other processing on the partial discharge signal data;
the diagnosis and identification module is connected with the data processing module, mainly comprises a trained multi-GRU model (namely a target fault diagnosis model), and is used for transmitting the data processed by the data processing module into the diagnosis and identification module and outputting a fault diagnosis result;
and the evaluation module is connected with the diagnosis and identification module and used for receiving the fault diagnosis result output by the diagnosis and identification module and providing reasonable processing measures by combining other monitoring information of the switch cabinet, such as temperature infrared and the like.
The system can be understood as an LMD-multi-GRU-based partial discharge fault diagnosis system for the switch cabinet, the partial discharge state of the switch cabinet can be continuously and stably monitored, reasonable processing measures are provided according to analysis, manual field intervention is avoided in the whole process, and safety of power station workers is guaranteed.
In this embodiment, a failure diagnosis processing apparatus for a switch cabinet is further provided, and the apparatus is used to implement the foregoing embodiments and preferred embodiments, and the description already made is omitted. As used hereinafter, the terms "module" and "apparatus" may refer to a combination of software and/or hardware that performs a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated.
According to an embodiment of the present invention, there is further provided an embodiment of an apparatus for implementing the method for diagnosing and processing a fault of a switch cabinet, and fig. 4 is a schematic structural diagram of an apparatus for diagnosing and processing a fault of a switch cabinet according to an embodiment of the present invention, and as shown in fig. 4, the apparatus for diagnosing and processing a fault of a switch cabinet includes: a first obtaining module 400, a processing module 402, a second obtaining module 404, and a training module 406, wherein:
the first obtaining module 400 is configured to obtain initial partial discharge fault data of the target switch cabinet;
the processing module 402 is connected to the first obtaining module 400, and configured to perform preprocessing on the initial partial discharge fault data by using a local mean decomposition algorithm to obtain first partial discharge fault data;
the second obtaining module 404 is connected to the processing module 402, and configured to obtain an initial fault diagnosis model constructed based on a multiplicative gated recurrent neural network algorithm;
the training module 406 is connected to the second obtaining module 404, and configured to input the first partial fault data to the initial fault diagnosis model for training, so as to obtain a target fault diagnosis model.
In the embodiment of the present invention, the first obtaining module 400 is configured to obtain initial partial discharge fault data of a target switch cabinet; the processing module 402 is connected to the first obtaining module 400, and configured to perform preprocessing on the initial partial discharge fault data by using a local mean decomposition algorithm to obtain first partial discharge fault data; the second obtaining module 404 is connected to the processing module 402, and configured to obtain an initial fault diagnosis model constructed based on a multiplicative gated recurrent neural network algorithm; the training module 406 is connected to the second obtaining module 404, and is configured to input the first partial fault data into the initial fault diagnosis model for training, so as to obtain a target fault diagnosis model, thereby achieving the purposes of improving the accuracy of obtaining the fault diagnosis model of the switch cabinet and accurately identifying the fault of the switch cabinet, and further achieving the technical effects of optimizing the fault diagnosis model of the switch cabinet, improving the fault diagnosis efficiency and the fault diagnosis accuracy of the switch cabinet, and further solving the technical problems of low fault diagnosis efficiency and poor diagnosis accuracy existing in the related art.
In an optional embodiment, the apparatus further comprises: the third acquisition module is used for acquiring data to be diagnosed corresponding to the switch cabinet to be detected; and the test module is used for inputting the data to be diagnosed to the target fault diagnosis model for testing to obtain a fault diagnosis result.
It should be noted that the above modules may be implemented by software or hardware, for example, for the latter, the following may be implemented: the modules can be located in the same processor; alternatively, the modules may be located in different processors in any combination.
It should be noted that the first obtaining module 400, the processing module 402, the second obtaining module 404, and the training module 406 correspond to steps S102 to S108 in the embodiment, and the modules are the same as the corresponding steps in the implementation example and the application scenario, but are not limited to the disclosure in the embodiment. It should be noted that the modules described above may be implemented in a computer terminal as part of an apparatus.
It should be noted that, for alternative or preferred embodiments of the present embodiment, reference may be made to the relevant description in the embodiments, and details are not described herein again.
The switch cabinet fault diagnosis processing apparatus may further include a processor and a memory, where the first obtaining module 400, the processing module 402, the second obtaining module 404, the training module 406, and the like are stored in the memory as program modules, and the processor executes the program modules stored in the memory to implement corresponding functions.
The processor comprises a kernel, and the kernel calls corresponding program modules from the memory, and the kernel can be set to be one or more than one. The memory may include volatile memory in a computer readable medium, random Access Memory (RAM) and/or nonvolatile memory such as Read Only Memory (ROM) or flash memory (flash RAM), and the memory includes at least one memory chip.
According to an embodiment of the present application, there is also provided an embodiment of a non-volatile storage medium. Optionally, in this embodiment, the nonvolatile storage medium includes a stored program, and when the program runs, the apparatus where the nonvolatile storage medium is located is controlled to execute any one of the switch cabinet fault diagnosis processing methods.
Optionally, in this embodiment, the nonvolatile storage medium may be located in any one of computer terminals in a computer terminal group in a computer network, or in any one of mobile terminals in a mobile terminal group, and the nonvolatile storage medium includes a stored program.
Optionally, the device in which the non-volatile storage medium is controlled to execute the following functions when the program runs: acquiring initial partial discharge fault data of a target switch cabinet; preprocessing the initial partial discharge fault data by adopting a local mean decomposition algorithm to obtain first partial discharge fault data; acquiring an initial fault diagnosis model constructed based on a multiplication gated cyclic neural network algorithm; and inputting the first partial fault data into the initial fault diagnosis model for training to obtain a target fault diagnosis model.
According to the embodiment of the application, the embodiment of the processor is also provided. Optionally, in this embodiment, the processor is configured to execute a program, where the program executes any one of the switch cabinet fault diagnosis processing methods when running.
There is further provided, according to an embodiment of the present application, an embodiment of a computer program product, which, when being executed on a data processing device, is adapted to execute a program initializing the steps of the switchgear cabinet fault diagnosis processing method of any of the above.
Optionally, the computer program product is adapted to perform a program for initializing the following method steps when executed on a data processing device: acquiring initial partial discharge fault data of a target switch cabinet; preprocessing the initial partial discharge fault data by adopting a local mean decomposition algorithm to obtain first partial discharge fault data; acquiring an initial fault diagnosis model constructed based on a multiplicative gated recurrent neural network algorithm; and inputting the first partial fault data into the initial fault diagnosis model for training to obtain a target fault diagnosis model.
The embodiment of the invention provides electronic equipment, which comprises a processor, a memory and a program which is stored on the memory and can run on the processor, wherein the processor executes the program and realizes the following steps: acquiring initial partial discharge fault data of a target switch cabinet; preprocessing the initial partial discharge fault data by adopting a local mean decomposition algorithm to obtain first partial discharge fault data; acquiring an initial fault diagnosis model constructed based on a multiplication gated cyclic neural network algorithm; and inputting the first partial fault data into the initial fault diagnosis model for training to obtain a target fault diagnosis model.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
In the above embodiments of the present invention, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.
In the embodiments provided in the present application, it should be understood that the disclosed technology can be implemented in other ways. The above-described apparatus embodiments are merely illustrative, and for example, the division of the above modules may be a logical division, and in actual implementation, there may be another division, for example, multiple modules or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, modules or indirect coupling or communication connection of modules, and may be in an electrical or other form.
The modules described as separate parts may or may not be physically separate, and parts displayed as modules may or may not be physical modules, may be located in one position, or may be distributed on a plurality of modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment.
In addition, functional modules in the embodiments of the present invention may be integrated into one processing module, or each of the modules may exist alone physically, or two or more modules are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode.
The integrated module may be stored in a computer-readable nonvolatile storage medium if it is implemented in the form of a software functional module and sold or used as a separate product. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a non-volatile storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned nonvolatile storage medium includes: a U-disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a removable hard disk, a magnetic or optical disk, and other various media capable of storing program codes.
The above is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, a plurality of modifications and embellishments can be made without departing from the principle of the present invention, and these modifications and embellishments should also be regarded as the protection scope of the present invention.
Claims (10)
1. A fault diagnosis processing method for a switch cabinet is characterized by comprising the following steps:
acquiring initial partial discharge fault data of a target switch cabinet;
preprocessing the initial partial discharge fault data by adopting a local mean decomposition algorithm to obtain first partial discharge fault data;
acquiring an initial fault diagnosis model constructed based on a multiplication gated cyclic neural network algorithm;
and inputting the first local fault data to the initial fault diagnosis model for training to obtain a target fault diagnosis model.
2. The method according to claim 1, wherein the preprocessing the initial partial discharge fault data by using a local mean decomposition algorithm to obtain first partial discharge fault data comprises:
performing first noise reduction processing on the initial partial discharge fault data by adopting the local mean decomposition algorithm to obtain second partial discharge fault data;
performing second noise reduction processing on the initial partial discharge fault data by adopting a singular value decomposition algorithm to obtain third partial discharge fault data;
and performing feature extraction processing on the third partial discharge fault data to obtain the first partial discharge fault data.
3. The method according to claim 2, wherein the performing a first noise reduction process on the initial partial discharge fault data by using the local mean decomposition algorithm to obtain a second partial discharge fault data includes:
obtaining a plurality of product functions based on the local mean decomposition algorithm;
taking the initial partial discharge fault data as original signals, and determining correlation coefficients between the multiple product functions and the original signals respectively;
determining a boundary position between a noise signal and a valid signal included in the original signal based on correlation coefficients between the plurality of product functions and the original signal, respectively;
and removing the signals before the boundary position in the original signals, and taking the removed original signals as the second partial discharge fault data.
4. The method according to claim 2, wherein the performing feature extraction processing on the third partial discharge fault data to obtain a feature extraction result as the first partial discharge fault data includes:
normalizing the third partial discharge fault data to obtain fourth partial discharge fault data;
and performing feature extraction processing on the fourth partial discharge fault data to obtain the first partial discharge fault data.
5. The method of claim 1, wherein inputting the first partial fault data into the initial fault diagnosis model for training to obtain a target fault diagnosis model comprises:
dividing the first partial release fault data into training set data and test set data;
inputting the training set data into the initial fault diagnosis model for training to obtain a trained initial fault diagnosis model;
inputting the test set data into the trained initial fault diagnosis model for testing to obtain a test result;
and taking the trained initial fault diagnosis model as the target fault diagnosis model under the condition that the test result meets a preset test condition.
6. The method according to any one of claims 1 to 5, wherein after the inputting the first partial fault data into the initial fault diagnosis model for training, obtaining a target fault diagnosis model, the method further comprises:
acquiring data to be diagnosed corresponding to a switch cabinet to be detected;
and inputting the data to be diagnosed to the target fault diagnosis model for testing to obtain a fault diagnosis result.
7. A switch cabinet fault diagnosis processing apparatus, comprising:
the first acquisition module is used for acquiring initial partial discharge fault data of the target switch cabinet;
the processing module is used for preprocessing the initial partial discharge fault data by adopting a local mean decomposition algorithm to obtain first partial discharge fault data;
the second acquisition module is used for acquiring an initial fault diagnosis model constructed based on a multiplication gated recurrent neural network algorithm;
and the training module is used for inputting the first partial fault data to the initial fault diagnosis model for training to obtain a target fault diagnosis model.
8. The apparatus of claim 7, further comprising:
the third acquisition module is used for acquiring data to be diagnosed corresponding to the switch cabinet to be detected;
and the test module is used for inputting the data to be diagnosed to the target fault diagnosis model for testing to obtain a fault diagnosis result.
9. A non-volatile storage medium, characterized in that it stores a plurality of instructions adapted to be loaded by a processor and to execute the method of fault diagnosis processing of a switchgear cabinet according to any one of claims 1 to 6.
10. An electronic device comprising one or more processors and memory for storing one or more programs, wherein the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the switchgear fault diagnostic processing method of any of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211732267.0A CN115902557A (en) | 2022-12-30 | 2022-12-30 | Switch cabinet fault diagnosis processing method and device and nonvolatile storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211732267.0A CN115902557A (en) | 2022-12-30 | 2022-12-30 | Switch cabinet fault diagnosis processing method and device and nonvolatile storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115902557A true CN115902557A (en) | 2023-04-04 |
Family
ID=86488276
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211732267.0A Pending CN115902557A (en) | 2022-12-30 | 2022-12-30 | Switch cabinet fault diagnosis processing method and device and nonvolatile storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115902557A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117590142A (en) * | 2024-01-19 | 2024-02-23 | 杭州万禾电力科技有限公司 | Switch cabinet fault diagnosis method and system based on deep learning |
-
2022
- 2022-12-30 CN CN202211732267.0A patent/CN115902557A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117590142A (en) * | 2024-01-19 | 2024-02-23 | 杭州万禾电力科技有限公司 | Switch cabinet fault diagnosis method and system based on deep learning |
| CN117590142B (en) * | 2024-01-19 | 2024-03-22 | 杭州万禾电力科技有限公司 | Switch cabinet fault diagnosis method and system based on deep learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ciancetta et al. | A new convolutional neural network-based system for NILM applications | |
| CN103558529B (en) | A kind of mode identification method of three-phase cartridge type supertension GIS partial discharge altogether | |
| CN113805064B (en) | Lithium ion battery pack health state prediction method based on deep learning | |
| CN103076547B (en) | Method for identifying GIS (Gas Insulated Switchgear) local discharge fault type mode based on support vector machines | |
| CN110829417B (en) | Electric power system transient stability prediction method based on LSTM double-structure model | |
| CN111564842B (en) | Method for statistical estimation of medium-and-long-term energy consumption in non-invasive electric load monitoring | |
| CN111965476B (en) | Low-voltage diagnosis method based on graph convolution neural network | |
| CN109100627A (en) | A kind of power equipment partial discharges fault diagnostic method based on end-to-end mode | |
| CN108562821B (en) | Method and system for determining single-phase earth fault line selection of power distribution network based on Softmax | |
| CN103439631A (en) | Method and system for detecting corrosion state of grounding grid | |
| CN116401532B (en) | Method and system for recognizing frequency instability of power system after disturbance | |
| CN115902557A (en) | Switch cabinet fault diagnosis processing method and device and nonvolatile storage medium | |
| CN113379252A (en) | Bayesian network high-voltage switch cabinet health system evaluation method based on multi-scale arrangement entropy | |
| CN115587309A (en) | Method, device and equipment for extracting key features of short-circuit resistance of transformer | |
| CN112085111A (en) | Load identification method and device | |
| CN116089882A (en) | Cable fault prediction processing method and device and electronic equipment | |
| CN112595918A (en) | Low-voltage meter reading fault detection method and device | |
| CN117291046A (en) | Management method and system for reliability test of battery BMS based on big data | |
| CN117471252A (en) | DC fault arc diagnosis method based on ICEEMDAN-KPCA feature extraction and SOA-KELM | |
| CN113740671A (en) | Fault arc identification method based on VMD and ELM | |
| CN115293303B (en) | High-voltage transmission line network monitoring method, system, equipment and medium | |
| CN111025100A (en) | Transformer ultrahigh frequency partial discharge signal mode identification method and device | |
| CN114397521A (en) | Fault diagnosis method and system for electronic equipment | |
| CN114236415A (en) | Power battery health state detection method, device, equipment and storage medium | |
| CN114487643A (en) | On-spot handing-over of extra-high voltage GIL equipment is accepted and is synthesized test platform |
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 |