A Fault Diagnosis Method, System and Medium for High-Voltage Bushing Based on Multi-Sensors

Abstract

The present invention discloses a multi-sensor-based high-voltage bushing fault diagnosis method, system and medium, which acquires first data information, where the first data information is the signal data of each sensor collected in the high-voltage bushing; preprocesses the first data information to obtain second data information; uses an optimized BP neural network model to extract the characteristic data information of the second data information, and the characteristic data information includes characteristic data on several different sensors; uses the D-S evidence theory algorithm to calculate the weight values of different fault types under the same characteristic data to obtain several weight values; divides several of the weight values according to different fault types, and judges the faults occurring in the monitoring system according to the magnitude of the weight values under the same fault type; the beneficial effect of the present invention is that it can improve the accuracy of judging the fault types occurring in the high-voltage bushing and reduce the efficiency of judging the high-voltage bushing faults.

Description

Technical Field

[0001] The present invention relates to the technical field of casing fault judgment, and more specifically, to a high-voltage casing fault diagnosis method, system and medium based on multi-sensors. Background Art

[0002] High-voltage casings are key components of equipment such as transformers and current transformers in the power system. In recent years, for high-voltage casings, a variety of on-line monitoring devices have been developed, using different sensors to monitor the state of the casings. Different sensors have been developed for the temperature, pressure, partial discharge, insulation, hydrogen content, etc. of the casings. Through different sensors, signals such as the temperature, pressure, UHF, HFCT, dielectric loss, capacitance, and hydrogen content of the casings can be collected or calculated.

[0003] Due to the complexity of the casing structure, when the casing is in operation, there are multiple structural movements inside it and different state variables. Therefore, when using different sensors for on-line monitoring and fault diagnosis of the casing, it is very likely that the faults and state variables do not correspond one by one. A certain fault may correspond to multiple state variables, and a certain state variable may also be caused by multiple faults.

[0004] However, in the prior art, when diagnosing casing faults, it is usually judged by the data information of a single sensor. When judging by the information of a single sensor, the collected information may be fuzzy or contradictory, reducing the accuracy of judging the type of fault occurring in the high-voltage casing and increasing the efficiency of judging the high-voltage casing fault.

[0005] In view of this, the present application is specifically proposed. Summary of the Invention

[0006] The technical problem to be solved by the present invention is that in the prior art, when judging the faults of high-voltage casings by collecting the information data of a single sensor, the collected information may be fuzzy or contradictory, which will reduce the accuracy of judging the type of faults occurring in the high-voltage casings and increase the efficiency of judging the high-voltage casing faults. The purpose is to provide a high-voltage casing fault diagnosis method, system and medium based on multi-sensors, which can improve the accuracy of judging the type of faults occurring in the high-voltage casings and reduce the efficiency of judging the high-voltage casing faults.

[0007] The present invention is achieved by the following technical solutions:

[0008] A high-voltage casing fault diagnosis method based on multi-sensors, the method steps include:

[0009] Obtain first data information, where the first data information is the signal data of each sensor collected in the high-voltage casing;

[0010] Preprocess the first data information to obtain second data information;

[0011] Use an optimized BP neural network model to extract the characteristic data information of the second data information, where the characteristic data information includes characteristic data from several different sensors;

[0012] Use the D-S evidence theory algorithm to calculate the weight values of different fault types under the same characteristic data, and obtain several weight values;

[0013] Divide several of the weight values according to different fault types, and judge the faults occurring in the monitoring system according to the magnitudes of the weight values under the same fault type.

[0014] In the traditional fault judgment of the on-line monitoring system for high-voltage bushings, usually, the information data of a single sensor is collected, and the information data of the single sensor is analyzed to achieve the diagnosis of the faults of the on-line monitoring system for high-voltage bushings. However, when using this method, sometimes the information data of the single sensor collected is fuzzy and inaccurate or the collected data information is contradictory, thus reducing the accuracy of judging the fault types occurring in the high-voltage bushings and increasing the efficiency of judging the faults of the high-voltage bushings; The present invention provides a method for diagnosing faults of high-voltage bushings based on multi-sensors. By collecting and fusing the information data of various sensors and using the method of combining the BP neural network with the D-S evidence theory algorithm to process the data information to judge the faults to which the on-line monitoring system belongs, it can improve the accuracy of judging the fault types occurring in the high-voltage bushings and reduce the efficiency of judging the faults of the high-voltage bushings.

[0015] Preferably, the preprocessing is to perform denoising or interference removal processing on the first data information.

[0016] Preferably, the steps for constructing the optimized BP neural network model are:

[0017] Obtain historical data information, where the historical data information is the signal data of faults occurring in each sensor in the high-voltage bushing and the corresponding fault types;

[0018] Perform denoising and interference removal processing on the historical data information to obtain sub-historical data information;

[0019] Construct a BP neural network model, and train the BP neural network model with the sub-historical data information, and perform optimization iteration using the error function of the standard neural network to obtain an optimized BP neural network model.

[0020] Preferably, the specific steps for obtaining the weight values include:

[0021] Among the feature data information, select any one piece of feature data, and use the BPA decision-making method in D-S evidence theory to calculate the conflict difference of this feature data under different fault types, and obtain a number of conflict differences;

[0022] Use the normalization method and combine the cosine theorem in trigonometric functions to process a number of the conflict differences to obtain the weight value corresponding to this feature data;

[0023] Traverse the feature data information to obtain a number of weight values.

[0024] Preferably, the fault type is an overheat fault or a discharge fault.

[0025] Preferably, the specific expression of the error function is:

[0026]

[0027] E is the error function, d k is the target output value, o k is the actual output value.

[0028] Preferably, the specific expression of the conflict difference is:

[0029]

[0030]

[0031] is the degree of conflict, x t is the t-th focal element, is the conflict, x ti is the i-th focal element in the t-th group, x tj is the i-th focal element and j-th focal element in the t-th group, m(x ti ) is the credibility assignment.

[0032] Preferably, the specific expression of the weight value is:

[0033]

[0034]

[0035]

[0036]

[0037] is the conflict difference, H t is the exponential operation value, β t is the weight value, ω t is the weight value.

[0038] The present invention also provides a high-voltage bushing fault diagnosis system based on multi-sensors, including a data acquisition module, a preprocessing module, a feature data extraction module, a weight value calculation module, and a judgment module.

[0039] The data acquisition module is used to acquire first data information, where the first data information is the signal data of each sensor in the collected high-voltage bushing.

[0040] The preprocessing module is used to preprocess the first data information to obtain second data information.

[0041] The feature data extraction module is used to extract the feature data information of the second data information by using an optimized BP neural network model, and the feature data information includes the feature data on several different sensors.

[0042] The weight calculation module is used to calculate the weight values of different fault types under the same feature data by using the D-S evidence theory algorithm to obtain several weight values.

[0043] The judgment module is used to classify several said weight values according to different fault types, and judge the faults occurring in the monitoring system according to the magnitudes of the weight values under the same fault type.

[0044] The present invention also provides a computer storage medium, on which a calculation program is stored. When the computer program is executed by a processor, the above-mentioned fault diagnosis method is realized.

[0045] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0046] A high-voltage bushing fault diagnosis method, system and medium provided by an embodiment of the present invention can improve the accuracy of judging the fault types occurring in the high-voltage bushing and reduce the efficiency of judging the high-voltage bushing fault by collecting and fusing various sensor information data and processing the data information by using the BP neural network combined with the D-S evidence theory algorithm to judge the faults of the on-line monitoring system. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention, and therefore should not be regarded as limiting the scope. For those of ordinary skill in the art, other related drawings can also be obtained based on these drawings without creative efforts.

[0048] Figure 1 is a high-voltage bushing fault diagnosis model;

[0049] Figure 2 It is the structural model of a parallel multi-source information fusion system;

[0050] Figure 3 It is the structure of a feedforward neural network;

[0051] Figure 4 It is the process of multi-source information fusion. Specific implementation mode

[0052] To make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below in conjunction with embodiments and drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0053] In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present invention. However, it will be apparent to those of ordinary skill in the art that: These specific details do not have to be employed to practice the present invention. In other embodiments, well-known structures, circuits, materials or methods have not been specifically described in order to avoid obscuring the present invention.

[0054] Throughout the specification, references to "one embodiment", "an embodiment", "an example" or "an example" mean that a particular feature, structure, or characteristic described in connection with the embodiment or example is included in at least one embodiment of the present invention. Thus, the phrases "one embodiment", "an embodiment", "an example" or "an example" appearing throughout the specification do not necessarily all refer to the same embodiment or example. Additionally, the particular features, structures, or characteristics may be combined in any suitable combination and / or sub-combination in one or more embodiments or examples. Further, those of ordinary skill in the art should understand that the diagrams provided herein are for illustrative purposes only and are not necessarily drawn to scale. The term "and / or" used herein includes any and all combinations of one or more of the associated listed items.

[0055] In the description of the present invention, the orientation or positional relationship indicated by the terms "front", "rear", "left", "right", "upper", "lower", "vertical", "horizontal", "high", "low", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the protection scope of the present invention.

[0056] Embodiment 1

[0057] In the traditional fault judgment of the on-line monitoring system for high-voltage bushings, usually the information data of a single sensor is collected and analyzed to achieve the diagnosis of the faults of the on-line monitoring system for high-voltage bushings. However, when using this method, sometimes the information data of the single sensor collected is fuzzy and inaccurate or the collected data information is contradictory, thus reducing the accuracy of judging the fault types occurring in the high-voltage bushings and increasing the efficiency of fault judgment for the high-voltage bushings.

[0058] This embodiment discloses a fault diagnosis method for high-voltage bushings based on multi-sensors. By collecting and fusing the information data of various sensors and adopting the method of combining the BP neural network with the D-S evidence theory algorithm to process the data information, the faults to which the on-line monitoring system belongs are judged, which can improve the accuracy of judging the fault types occurring in the high-voltage bushings and reduce the efficiency of fault judgment for the high-voltage bushings. The schematic diagram of the specific diagnosis method in this embodiment is as Figures 1 to 4 shown, and the method steps include:

[0059] S1: Obtain the first data information, where the first data information is the signal data of each sensor in the collected high-voltage bushing;

[0060] In step S1, the high-voltage bushing is an outgoing line device that leads the high-voltage wire inside the transformer to the outside of the oil tank. It not only serves as the ground insulation of the lead wire but also plays a role in fixing the lead wire, and is one of the important accessories of the transformer; therefore, when the transformer is operating, the bushing is usually monitored online, and the monitoring is carried out through various sensors set. Therefore, in this embodiment, it is to obtain the data of the various sensors for online monitoring of the bushing, and fusing the data can increase the accuracy of judging the fault types.

[0061] S2: Preprocess the first data information to obtain the second data information; the preprocessing is to perform noise removal or interference removal processing on the first data information.

[0062] In step S2, there may be certain noise or interference information in the collected first data information, and the noise and interference information need to be removed, and the method used is filtering.

[0063] S3: Use the optimized BP neural network model to extract the characteristic data information of the second data information, where the characteristic data information includes the characteristic data on several different sensors;

[0064] The construction steps of the optimized BP neural network model are:

[0065] Obtain historical data information, where the historical data information is the signal data of faults occurring in each sensor in the high-voltage bushing and the corresponding fault types; perform denoising and interference removal processing on the historical data information to obtain sub-historical data information; construct a BP neural network model, and train the BP neural network model with the sub-historical data information, and use the error function of the standard neural network for optimization iteration to obtain an optimized BP neural network model.

[0066] By continuously optimizing and iteratively updating the BP neural network model with the collected historical data information, the accuracy of the model in processing relevant data information can be increased.

[0067] The specific BP neural network model is as follows:

[0068] The input layer, middle layer, and output layer of the standard BP network have Ni, Nj, and Nk neurons respectively. The input of the j-th neuron in the middle layer is: i i, N j and N k neurons. The input of the j-th neuron in the middle layer is:

[0069]

[0070] In the formula, ω ij is the weight from the i-th neuron in the input layer to the j-th neuron in the middle layer; o i is the output of the i-th neuron in the input layer. The input of the k-th neuron in the output layer is:

[0071]

[0072] In the formula, ω jk is the weight from the j-th neuron in the middle layer to the k-th neuron in the output layer; o j is the output of the j-th neuron in the middle layer.

[0073] The outputs of the input layer, middle layer, and output layer are respectively:

[0074] o i = net j = x i

[0075]

[0076]

[0077] θ j and θ k are the thresholds of the j-th neuron in the middle layer and the k-th neuron in the output layer respectively. x i is the signal collected by each sensor.

[0078] The training of the BP network adopts the γ learning law based on the gradient method, and its goal is to minimize the mean square error between the network output and the training samples. In the standard BP neural network, let the training sample be P, where the input vectors are x1, x2... xp; the output vectors are y1, y2... yp; the corresponding teacher value (sample) vectors are t1, t2... tp; then the mean square error of the P sample is:

[0079]

[0080] In the formula, tpk and ypk are the teacher value and the actual output value of the k-th output neuron for the p-th sample, respectively.

[0081] At this time, the weight adjustment of the middle layer is:

[0082] Δω ij (n + 1) = ηδ jp o jp +αΔw ij (n)

[0083]

[0084] At this time, the weight adjustment of the output layer is:

[0085] Δω jk (n + 1) = ηδ jp o jp +αΔw jk (n)

[0086] δ kp =(t kp -y kp )f k '(net kp )

[0087] In the formula, η is the learning rate and α is the momentum factor.

[0088] The specific expression of the error function is:

[0089]

[0090] E is the error function, d k is the target output value, o k is the actual output value.

[0091] The error function reflects the different "distance" scales between d k and o k The purpose of training the BP network is to make the actual output o k as close as possible to the target output d k, measured by this degree of proximity. Using this error function can effectively improve the problem of slow convergence speed during the training of the BP neural network and increase the training speed of the BP neural network.

[0092] After using this error function, the weight of the middle layer of the neural network is adjusted to:

[0093]

[0094] At this time, the weight of the output layer is adjusted to:

[0095]

[0096] Among them, ω ij is the weight from the i-th neuron in the input layer to the j-th neuron in the middle layer; ω jk is the weight from the j-th neuron in the middle layer to the k-th neuron in the output layer; η is the learning rate adjustment factor. O i is the output of the i-th neuron in the input layer. O j is the output of the K-th neuron in the middle layer.

[0097] S4: Adopt the D-S evidence theory algorithm to calculate the weight values of different fault types under the same characteristic data and obtain several weight values;

[0098] The specific steps for obtaining the weight values include:

[0099] In the characteristic data information, select any one characteristic data, and use the BPA decision method in the D-S evidence theory to calculate the conflict difference of this characteristic data under different fault types to obtain several conflict differences; the fault types are overheating faults or discharge faults.

[0100] Adopt the normalization method and combine the cosine theorem in trigonometric functions to process several of the conflict differences to obtain the weight value corresponding to this characteristic data;

[0101] Traverse the characteristic data information to obtain several weight values.

[0102] In this embodiment, a method for redistributing the evidence BPA is proposed for the D-S evidence theory algorithm, which effectively overcomes the influence of evidence conflict on information fusion and reasonably uses the method of evidence reasoning to numerically describe the certainty and hesitation in judgment.

[0103] Define m as the BPA function on the recognition space θ, A and B are the focal elements on the recognition space, and redefine the BPA on the recognition space using the formula.

[0104]

[0105] SetP(A) represents the degree of support of the basic probability assignment for each subset.

[0106]

[0107] The conflict degree Diff of the same target is expressed as:

[0108] Diff(A) = |SetP mi (A) - SetP mj (A)|

[0109] In this embodiment, the D-S evidence theory algorithm is improved, and the evidence is re-weighted according to the reliability of the evidence to obtain a better fusion result.

[0110] Let θ be the sample space, xi be the focal element in the sample space θ, i = 1, 2,... n, and there are N groups of evidence functions. Denote mi = {xt1, xt2, xt3.....xti} where i = 1, 2,... n, t = 1, 2,... N. Use βi to represent the weight, where i = 1, 2, 3....N.

[0111] First, redistribute the BPA of the evidence respectively.

[0112]

[0113] Secondly, calculate the conflict difference of the same focal element under different evidences.

[0114]

[0115] is the conflict degree, x t is the t-th group of focal elements, is the conflict, x ti is the i-th focal element in the t-th group, x tj is the i-th focal element and j-th focal element in the t-th group, m(x ti ) is the credibility assignment.

[0116] Thirdly, normalize the conflict difference.

[0117]

[0118] satisfies, is the conflict difference, H t is the exponential operation value, β t is the weight, ω t is the weight value.

[0119] Then calculate the normalized value.

[0120]

[0121] The weight value can be expressed as:

[0122]

[0123] Introduce the cosine theorem in trigonometric functions, fix the weight value between [0,1], and endow the weight value by using the curve characteristics of the cosine function, so that the obtained weight value is relatively smooth. That is:

[0124]

[0125] In this way, the weight vector composed of the weight coefficients of each evidence is determined.

[0126] S5: Divide several said weight values according to different fault types, and judge the faults occurring in the monitoring system according to the magnitudes of the weight values under the same fault type.

[0127] Combine the BP neural network and the D-S evidence theory to fuse the eigenvalue and eigenvector. Set the uncertain factor fi as the training error of the BP neural network, and normalize the output as the basic probability value of each focal element. Its calculation formula is:

[0128]

[0129] In the formula: fi represents the fault mode, i=(0,1,2,3,4,5,6), and y(fi) represents the output result of the BP neural network.

[0130] Among them, En is the network error of this sample, and tnj and ynj are the expected value and actual value corresponding to the jth neuron respectively.

[0131]

[0132] First, calculate the BP neural network to process the input values of each sensor to obtain the corresponding mass function values for different fault type confirmations, and then apply the D-S evidence theory rules to finally obtain the fusion result.

[0133] Specific implementation process:

[0134] Construct a BP diagnosis network. The number of neurons in the input layer of the BP network is 9 characteristic parameters, and the number of neurons in the output layer is the number of faults (representing different fault types), that is, the output objective function F = {f1, f2, f3, f4, f5, f6}. The expected output of the neural network is represented by [0, 1], where 1 represents the existence of a fault and 0 represents the non-existence of a fault. Table 1 shows the input samples of the neural network, which are 9 characteristic parameters such as the root mean square x1, standard deviation x2, temperature index x3, pressure index x4, oil level index x5, dielectric loss index x6, capacitance index x7, leakage current x8, and angular difference x9.

[0135] Table 1 Input samples of the neural network

[0136]

[0137] According to the measured point data in Table 1, first perform local diagnosis using the BP neural network. Set the input layer of the BP network to 6 nodes, the hidden layer to 17 nodes, and the output layer to 6 nodes, which represent six kinds of faults respectively. Use the training set to train the three-layer BP network, and then extract some samples from each fault sample for training. The training results are shown in Table 2:

[0138] Table 2 Training results of the BP neural network

[0139]

[0140]

[0141] By combining the BP neural network with the D-S evidence theory using the present invention, the fusion result of the samples can be obtained by applying the rules of the D-S evidence theory, as shown in Table 3.

[0142] Table 3 Fusion results of the D-S evidence theory

[0143]

[0144] By comparing Table 2 and Table 3, it can be seen that: through comprehensive diagnosis of information fusion, the diagnosis accuracy is greatly improved. There are not very ideal results in each sample. If the decision is made using the data of a single measured point, it is easy to misjudge. However, by using the multi-sensor information fusion method, considering multiple parameters and variables comprehensively, first performing feature layer fusion using the BP neural network and decision layer fusion using the D-S synthesis rule, the finally obtained results are relatively ideal, and all the confidence degrees are above 0.95, thus proving the effectiveness of the information fusion casing fault intelligent diagnosis method combining the BP neural network and the D-S evidence theory.

[0145] A fault diagnosis method for an on-line monitoring system of a high-voltage bushing provided in this embodiment processes data information by collecting and fusing various sensor information data and using a method combining a BP neural network and a D-S evidence theory algorithm to determine the faults of the on-line monitoring system, which can improve the accuracy of judging the types of faults occurring in the high-voltage bushing and reduce the efficiency of judging the faults of the high-voltage bushing.

[0146] Embodiment 2

[0147] This embodiment discloses a high-voltage bushing fault diagnosis system based on multi-sensors. This embodiment is to implement the fault diagnosis method in Embodiment 1, including a data acquisition module, a preprocessing module, a feature data extraction module, a weight value calculation module, and a judgment module.

[0148] The data acquisition module is used to acquire first data information, and the first data information is the signal data of each sensor in the collected high-voltage bushing.

[0149] The preprocessing module is used to preprocess the first data information to obtain second data information.

[0150] The feature data extraction module is used to extract the feature data information of the second data information by using an optimized BP neural network model, and the feature data information includes the feature data on several different sensors.

[0151] The weight calculation module is used to calculate the weight values of different fault types under the same feature data by using the D-S evidence theory algorithm to obtain several weight values.

[0152] The judgment module is used to classify several of the weight values according to different fault types and judge the faults occurring in the monitoring system according to the magnitudes of the weight values under the same fault type.

[0153] Embodiment 3

[0154] This embodiment discloses a computer storage medium on which a computing program is stored. When the computer program is executed by a processor, it implements the fault diagnosis method described in Embodiment 1.

[0155] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0156] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program issued instructions. These computer program issued instructions can be provided to the processor of a general purpose computer, a special purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the issued instructions executed by the processor of the computer or other programmable data processing devices generate means for realizing the functions specified in the Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0157] These computer program issued instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the issued instructions stored in the computer-readable memory generate a manufactured article including the issued instruction means, and the issued instruction means realizes the functions specified in the Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0158] These computer program issued instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the issued instructions executed on the computer or other programmable device provide steps for realizing the functions specified in the Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0159] The specific embodiments described above further elaborate on the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above is only the specific embodiments of the present invention and is not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A multi-sensor based high voltage bushing fault diagnosis method, characterized in that, The method steps include: Obtain first data information, where the first data information is the signal data of each sensor in the high-voltage bushing collected; Preprocess the first data information to obtain second data information; Use an optimized BP neural network model to extract the characteristic data information of the second data information, where the characteristic data information includes characteristic data on several different sensors; Use the D-S evidence theory algorithm to calculate the weight values of different fault types under the same characteristic data to obtain several weight values; Classify several of the weight values according to different fault types, and judge the faults occurring in the monitoring system according to the magnitudes of the weight values under the same fault type; The construction steps of the optimized BP neural network model are as follows: Obtain historical data information, where the historical data information is the signal data of each sensor in the high-voltage bushing when a fault occurs and the corresponding fault type; Denoise and anti-interference process the historical data information to obtain sub-historical data information; Construct a BP neural network model, and train the BP neural network model through the sub-historical data information, and perform optimization iteration using the improved error function of the neural network to obtain an optimized BP neural network model; The specific steps for obtaining the weight values include: In the characteristic data information, select any one characteristic data, use the BPA decision method in the D-S evidence theory to calculate the conflict differences of this characteristic data under different fault types to obtain several conflict differences; Use the normalization method and combine the cosine theorem in trigonometric functions to process several of the conflict differences to obtain the weight value corresponding to this characteristic data; Traverse the characteristic data information to obtain several weight values; The specific expression of the error function is: E is the error function, d k is the target output value, o k is the actual output value; The specific expression of the conflict difference is: is the conflict degree, x t is the t-th group of focal elements, is the conflict, x ti is the i-th focal element in the t-th group, x tj is the i-th and j-th focal elements in the t-th group, m(x ti ) is the credibility assignment; The specific expression of the weight value is: α xt is the conflict difference, H t is the exponential operation value, β t is the weight value, ω t is the weight.

2. The multi-sensor based high voltage bushing fault diagnosis method according to claim 1, characterized in that, The preprocessing is to denoise or anti-interference process the first data information.

3. The multi-sensor based high voltage bushing fault diagnosis method according to claim 1, characterized in that, The fault types are overheating faults or discharge faults.

4. A multi-sensor based high voltage bushing fault diagnosis system, characterized in that, Used to implement the fault diagnosis method described in claim 1, including a data acquisition module, a preprocessing module, a characteristic data extraction module, a weight value calculation module, and a judgment module. The data acquisition module is used to obtain first data information, where the first data information is the signal data of each sensor in the high-voltage bushing collected; The preprocessing module is used to preprocess the first data information to obtain second data information; The characteristic data extraction module is used to use an optimized BP neural network model to extract the characteristic data information of the second data information, where the characteristic data information includes characteristic data on several different sensors; The weight value calculation module is used to use the D-S evidence theory algorithm to calculate the weight values of different fault types under the same characteristic data to obtain several weight values; The judgment module is used to classify several of the weight values according to different fault types, and judge the faults occurring in the monitoring system according to the magnitudes of the weight values under the same fault type.

5. A computer storage medium, on which a computing program is stored, characterized in that, When this computer program is executed by a processor, it implements the fault diagnosis method described in any one of claims 1 to 3.

Images