Cable early fault early warning method and system based on multi-feature fusion and iterative prediction model

Abstract

The invention discloses a cable early fault early warning method and system based on multi-feature fusion and an iterative prediction model, and relates to the technical field of wire and cable health management, and the method comprises the steps: carrying out the standardization processing of the time series data of a plurality of early fault features, screening out important features through a gray correlation analysis method, and carrying out the calculation of the early fault features; the weights of the multiple features are calculated through a subjective and objective comprehensive weight method, and the multiple features are fused into a comprehensive feature; constructing an augmented support vector regression model; training an augmented support vector regression model by using the known comprehensive features, and carrying out iterative prediction on unknown comprehensive features; and determining the residual service life of the cable according to the comprehensive characteristic threshold value corresponding to the cable failure and the iterative prediction result, and setting a residual service life early warning threshold value to complete early fault early warning. According to the method, the problems of insufficient data utilization and low feature change prediction accuracy in the early fault early warning process of the cable are solved, and the method is of great significance for guaranteeing safe operation of a cable system.

Description

Technical Field

[0001] The present invention relates to the technical field of wire and cable health management, and more specifically to a cable early fault warning method and system based on multi-feature fusion and iterative prediction model. Background Art

[0002] As a key component of power transmission and distribution, cables play a vital role in many fields such as energy, transportation, and communications. Before a permanent failure occurs in a cable, there will be several to dozens of self-extinguishing arc discharges, which are called early faults. The intermittent arcing frequency and insulation degradation of early faults promote each other and eventually develop into permanent faults. Therefore, early warning of cable failures is of great significance to ensure the stability and safety of the power system.

[0003] Therefore, how to propose a cable early fault warning method and system based on multi-feature fusion and iterative prediction model, collect and fuse multi-source data of cable early fault characteristics, and use data-driven machine learning algorithms to build a more accurate early fault feature iterative prediction model to complete early fault warning is a problem that technical personnel in this field urgently need to solve. Summary of the invention

[0004] In view of this, the present invention provides a cable early fault warning method and system based on multi-feature fusion and iterative prediction model, by collecting and fusing multi-source data of cable early fault features, using data-driven machine learning algorithm to build a more accurate early fault feature iterative prediction model, to complete early fault warning. In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0005] A cable early fault warning method based on multi-feature fusion and iterative prediction model, comprising:

[0006] Standardize the time series data of early fault characteristics and filter out important features;

[0007] Calculate the weights of multiple important features and merge them into comprehensive features;

[0008] Combining the Chaos Grey Wolf Optimization-Cuckoo Search Hybrid Algorithm with the Support Vector Regression Model, an Augmented Support Vector Regression Model is constructed.

[0009] The augmented support vector regression model is trained using comprehensive features, and the data to be tested is iteratively predicted based on the trained augmented support vector regression model;

[0010] Early fault warning is carried out through the comprehensive feature threshold corresponding to cable failure and the iterative prediction results.

[0011] Optionally, the performing of the standardization process to screen out important features includes: respectively calculating early fault features with a decreasing trend and early fault features with an increasing trend.

[0012] Optionally, the standardization process includes:

[0013]

[0014] In the formula, H d is an early fault feature with a decreasing trend, H i is an early fault feature with an increasing trend, H t is the early fault characteristic value at time t, H0 is the initial early fault characteristic, H crit It is the key value of early fault characteristics.

[0015] Optionally, the calculating weights of multiple important features and fusing the multiple important features into a comprehensive feature includes: calculating the weights of multiple important features by a subjective and objective comprehensive weight method, wherein the subjective and objective comprehensive weight method is obtained by combining the fuzzy hierarchical analysis method and the entropy weight method, obtaining a comprehensive weight based on the calculated weights of multiple important features, and calculating the comprehensive feature based on the comprehensive weight.

[0016] Optionally, the fusion into comprehensive features includes:

[0017]

[0018]

[0019] In the formula, α k is the weight corresponding to the early fault feature calculated based on the fuzzy analytic hierarchy process, β k is the weight corresponding to the early fault feature calculated based on the entropy weight method, f ij is the element in the fuzzy consistent matrix, n is the number of early fault features, E k is the entropy value of early fault characteristics, w k is the comprehensive weight corresponding to the kth early fault feature;

[0020] H c =H1w1+H2w2+…+H k w k ;

[0021] In the formula, H c is a comprehensive feature, H k is the kth early fault feature.

[0022] Optionally, the chaotic gray wolf optimization-cuckoo search hybrid algorithm includes: updating the initial position of the gray wolf individual; updating the position of the gray wolf individual again through the bird's nest position in the cuckoo search algorithm; and determining the updated position of the ordinary gray wolf based on the optimal solution for updating the position of the gray wolf individual again.

[0023] Optionally, determining the updated position of the common gray wolf according to the optimal solution for re-updating the position of the individual gray wolf comprises:

[0024] Update the initial positions of the gray wolf individuals (α, β, and δ):

[0025] X1=X α (t)-A1|C1X α (t)-mX(t)|;

[0026] X2=X β (t)-A2|C2X β (t)-mX(t)|;

[0027] X3=X δ (t)-A3|C3X δ (t)-mX(t)|;

[0028] Among them, X α , X β , X δ are the position vectors of α wolf, β wolf and δ wolf respectively, m is the Tent chaotic variable, A1, A2, A3, C1, C2 and C3 are coefficient vectors, t is the current iteration number, X1, X2 and X3 are the first three optimal solutions;

[0029]

[0030] The position of the individual gray wolf is updated again through the updating formula of the bird nest position in the cuckoo search algorithm;

[0031]

[0032] Where i is the number of individuals in the current wolf pack, τ is the step length control parameter, S is the Lévy flight search step length, and x n For the location of the bird's nest;

[0033] Finally, the updated position of the common gray wolf is obtained according to the formula;

[0034] X(t+1)=(X1+X2+X3) / 3;

[0035] Where X is the updated position of the common gray wolf.

[0036] Optionally, the constructing of an augmented support vector regression model includes:

[0037] S1: Set the initial division ratio, divide the acquired early fault feature time series into two sections, and obtain the front and back time series data sets;

[0038] S2: Set the initial time scale, kernel scale parameters and box constraint parameters, and use the previous time series to train the support vector regression model;

[0039] S3: Use the trained support vector regression model to iteratively predict the data of the next period of time, and use the root mean square error between the predicted value and the true value of the next period of time as the objective function, and use the chaotic gray wolf optimization-cuckoo search hybrid algorithm to determine l optimal parameters to be determined;

[0040] S4: Substitute the l best unknown parameters into the support vector regression model to establish an augmented support vector regression model.

[0041] Optionally, the iterative prediction of the test data based on the trained augmented support vector regression model includes: when the time scale of the input variable in the iterative prediction process is p, it means that the previous p time series points are used to predict the next time series point, and the predicted value and its previous p-1 time series points are reorganized into a new time series group to make a prediction for the next moment; iteratively updating p time series until the prediction of the qth time series point is completed.

[0042] Optionally, a cable early fault warning system based on multi-feature fusion and iterative prediction model, comprising:

[0043] Standardization processing module: used to standardize the time series data of early fault characteristics and filter out important features;

[0044] Fusion module: used to calculate the weights of multiple important features and fuse multiple important features into comprehensive features;

[0045] Model building module: used to combine the chaotic grey wolf optimization-cuckoo search hybrid algorithm with the support vector regression model to build an augmented support vector regression model;

[0046] Iterative prediction module: used to train the augmented support vector regression model using comprehensive features, and iteratively predict the test data based on the trained augmented support vector regression model;

[0047] Fault warning module: used to provide early fault warning through the comprehensive feature threshold and iterative prediction results corresponding to cable failure.

[0048] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a cable early fault warning method and system based on multi-feature fusion and iterative prediction model, which has the following beneficial effects:

[0049] The present invention proposes a cable early fault warning method based on multi-feature fusion and iterative prediction model, including: standardizing the time series data of early fault features to screen out important features; calculating the weights of multiple important features and fusing multiple important features into comprehensive features; combining the chaotic gray wolf optimization-cuckoo search hybrid algorithm with the support vector regression model to construct an augmented support vector regression model; using the comprehensive features to train the augmented support vector regression model, and iteratively predicting the test data according to the trained augmented support vector regression model; and performing early fault warning by the comprehensive feature threshold and iterative prediction results corresponding to the cable failure. The present invention provides a cable early fault warning method and system based on multi-feature fusion and iterative prediction model, which adopts multi-feature fusion to comprehensively consider the electrical features, thermal features and optical features of the cable early fault, thereby realizing reliable early warning of the cable early fault; by establishing an early fault feature iterative prediction model, it is possible to accurately predict the change trend of the early fault features using known feature data, so that the cable early fault warning result is more accurate; reliable and accurate early fault warning is conducive to formulating more reasonable maintenance plans and replacement strategies, thereby avoiding power outages and economic losses caused by cable failures. By collecting and integrating multi-source data on early fault characteristics of cables, a more accurate iterative prediction model of early fault characteristics is constructed using data-driven machine learning algorithms. By iteratively optimizing model parameters, the accuracy and robustness of the prediction can be improved, while reliable early fault warning can be achieved, providing a scientific basis for cable maintenance and management. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0051] Figure 1 A schematic flow chart of a cable early fault warning method based on multi-feature fusion and iterative prediction model provided by the present invention.

[0052] Figure 2 A curve diagram showing the changes in the comprehensive characteristics provided by the present invention.

[0053] Figure 3 A schematic diagram of the comprehensive feature iterative prediction process based on the augmented support vector regression model provided by the present invention.

[0054] Figure 4 This is a graph of the iterative prediction of the comprehensive features of the three model outputs provided by the present invention.

[0055] Figure 5 This is a diagram showing the prediction results of the remaining service life of the cable under different operating years provided by the present invention. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] The embodiment of the present invention discloses a cable early fault warning method based on multi-feature fusion and iterative prediction model, comprising:

[0058] Standardize the time series data of early fault characteristics and filter out important features;

[0059] Calculate the weights of multiple important features and merge them into comprehensive features;

[0060] Combining the Chaos Grey Wolf Optimization-Cuckoo Search Hybrid Algorithm with the Support Vector Regression Model, an Augmented Support Vector Regression Model is constructed.

[0061] The augmented support vector regression model is trained using comprehensive features, and the data to be tested is iteratively predicted based on the trained augmented support vector regression model;

[0062] Early fault warning is carried out through the comprehensive feature threshold corresponding to cable failure and the iterative prediction results.

[0063] In a specific implementation, a cable early fault warning method based on multi-feature fusion and iterative prediction model specifically includes the following steps:

[0064] The time series data of multiple early fault features are standardized, the important features are screened out using the grey correlation analysis method, and the weights of multiple features are calculated using the subjective and objective comprehensive weight method, and then the multiple features are integrated into comprehensive features;

[0065] The Chaos Grey Wolf Optimization-Cuckoo Search hybrid algorithm is combined with the support vector regression model to form an augmented support vector regression model;

[0066] The augmented support vector regression model is trained using known comprehensive features, and the unknown comprehensive features are iteratively predicted;

[0067] The remaining service life of the cable is determined by the comprehensive feature threshold corresponding to the cable failure and the iterative prediction results, and the remaining service life warning threshold is set to complete early fault warning.

[0068] In a specific embodiment, during the standardization process, for early fault features with a decreasing trend, the transformation formula is as follows (1). For early fault features with an increasing trend, the transformation formula is as follows (2). The subjective and objective comprehensive weight method is obtained by combining the fuzzy hierarchical analysis method and the entropy weight method, and the comprehensive weight calculation formula is as follows (3), and the calculation formula of the comprehensive feature is as follows (6).

[0069]

[0070] In the formula, H d is an early fault feature with a decreasing trend, H i is an early fault feature with an increasing trend, H t is the early fault characteristic value at time t, H0 is the initial early fault characteristic, H crit It is the key value of early fault characteristics.

[0071]

[0072] In the formula, α k is the weight corresponding to the early fault feature calculated based on the fuzzy analytic hierarchy process, β k is the weight corresponding to the early fault feature calculated based on the entropy weight method, f ij is the element in the fuzzy consistent matrix, n is the number of early fault features, E k is the entropy value of the early fault feature.

[0073] H c =H1w1+H2w2+…+H k w k (6)

[0074] In the formula, H c is a comprehensive feature, H k is the kth early fault feature, w k is the comprehensive weight corresponding to the kth early fault feature.

[0075] In a specific embodiment, the initial position of the gray wolf individuals (α, β and δ) in the chaotic gray wolf optimization-cuckoo search hybrid algorithm is first updated by formula (7)-formula (9), wherein the parameter m is the Tent chaotic variable, calculated by formula (10). Then, the position of the gray wolf individuals is updated again by the update formula (11) of the bird nest position in the cuckoo search algorithm, and finally the updated position of the common gray wolf is obtained according to formula (12).

[0076] X1=X α (t)-A1|C1X α (t)-mX(t)| (7)

[0077] X2=X β (t)-A2|C2X β (t)-mX(t)| (8)

[0078] X3=X δ (t)-A3|C3X δ (t)-mX(t)| (9)

[0079]

[0080] X(t+1)=(X1+X2+X3) / 3 (12)

[0081] Where, X α , X β , X δ are the position vectors of α wolf, β wolf, and δ wolf respectively, m is the chaotic variable, A1, A2, A3, C1, and C2 are the coefficient vectors, t is the current iteration number, X1, X2, and X3 are the first three optimal solutions, i is the number of individuals in the current wolf pack, τ is the step length control parameter, S is the Lévy flight search step length, and x n is the location of the bird's nest, and X is the updated location of the common gray wolf.

[0082] In a specific embodiment, the specific steps of establishing the augmented support vector regression model are as follows:

[0083] 1) Set the initial division ratio and divide the acquired early fault feature time series into two sections to obtain the front and back time series data sets;

[0084] 2) Setting the initial time scale, kernel scale parameters and box constraint parameters, and using the previous time series to train the support vector regression model;

[0085] 3) Use the trained support vector regression model to iteratively predict the data of the next period of time, and use the root mean square error between the predicted value and the true value of the next period of time as the objective function, and use the chaotic gray wolf optimization-cuckoo search hybrid algorithm to determine the four optimal parameters;

[0086] 4) Substitute the four best unknown parameters into the support vector regression model to establish an augmented support vector regression model.

[0087] In a specific embodiment, the specific content of the iterative prediction is: when the time scale of the input variable in the iterative prediction process is p, it means that the previous p time series points are used to predict the next time series point, and the predicted value and its previous p-1 time series points are reorganized into a new time series group to predict the next moment. In this way, p time series are iteratively updated until the prediction of the qth time series point is completed.

[0088] Specifically, in the iterative prediction process based on the augmented support vector regression model, the time series of the comprehensive features is shown in formula (13). There are q variables in the time series, the first g variables are used to construct the training set, and the last qg variables are used as the test set. When the time scale of the input of the augmented support vector regression model is p, the input and output of the training set constructed by the first g variables are shown in formula (14) and formula (15), respectively.

[0089] H t =[H c1 H c3 H c4 …H cq ] (13)

[0090]

[0091] In the formula, H t is the time series of comprehensive features, X t is the input of the training set, Y t is the output of the training set.

[0092] In a specific implementation, a cable early fault warning method based on multi-feature fusion and iterative prediction model, such as Figure 1 As shown, the following steps are included:

[0093] Step S01: Taking the early fault warning of cross-linked polyethylene cables as an example, the early fault features with a decreasing trend are transformed according to formula (1), and the early fault features with an increasing trend are transformed according to formula (2), and multiple dimensionless features with the same starting point are obtained. The important features are screened out using the grey correlation analysis method. The subjective weight value is calculated according to formula (4), the objective weight value is calculated according to formula (5), and the subjective weight value and the objective weight value are combined according to formula (3) to calculate the comprehensive weight value of each feature. Multi-feature fusion is realized according to formula (6), and the comprehensive feature is calculated. The change curve of the comprehensive feature is shown as follows: Figure 2 shown.

[0094]

[0095] In the formula, H d is an early fault feature with a decreasing trend, H i is an early fault feature with an increasing trend, H t is the early fault characteristic value at time t, H0 is the initial early fault characteristic, H crit It is the key value of early fault characteristics.

[0096]

[0097] In the formula, α kis the weight corresponding to the early fault feature calculated based on the fuzzy analytic hierarchy process, β k is the weight corresponding to the early fault feature calculated based on the entropy weight method, f ij is the element in the fuzzy consistent matrix, n is the number of early fault features, E k is the entropy value of early fault characteristics, w k is the comprehensive weight corresponding to the kth early fault feature.

[0098] H c =H1w1+H2w2+…+H k w k (6)

[0099] In the formula, H c is a comprehensive feature, H k is the kth early fault feature.

[0100] Step S02: Figure 3 The figure is a schematic diagram of the iterative prediction process of comprehensive features based on the augmented support vector regression model. The support vector regression model is trained using the first segment of data in 60% of the known data set, and the second segment of data is used to establish the objective function, which is the root mean square error between the predicted value and the true value after the support vector regression model obtained by the preliminary training performs iterative prediction on the second segment of data. The time scale varies in the range of [1,10], the kernel scale parameter varies in the range of [1,1e3], the box constraint parameter varies in the range of [1,1e3], and the division ratio of the first segment of data and the second segment of data in the training set varies in the range of [0.4,0.65]. The optimal division ratio of the first segment of data and the second segment of data in the training set is determined to be 0.5183, the optimal time scale is 9, the optimal kernel scale parameter is 88, and the optimal box constraint parameter is 706 using the chaotic gray wolf optimization-cuckoo search hybrid algorithm. Substituting these optimal parameters into the support vector regression model, the augmented support vector regression model is obtained.

[0101] Based on the established augmented support vector regression model, particle filter-exponential model and long short-term memory neural network regression model, iterative prediction of comprehensive features is performed respectively. The prediction curve is as follows: Figure 4 As shown, the average absolute errors of the prediction results are 1.2469, 3.5236 and 5.4107 respectively. Comparing the prediction errors of the three models, it can be seen that the comprehensive feature iterative prediction model proposed in the present invention has the best performance.

[0102] Step S03: The comprehensive feature corresponding to the cable failure is calculated to be 66.63% based on the actual detection values ​​of the multiple features corresponding to the cable failure. When the comprehensive feature changes to the failure threshold of 66.63%, the remaining service life of the cable can be calculated based on the cable failure time and the current time. The prediction results of the remaining service life of the cable under different operating years are as follows: Figure 5 As shown. The prediction performance of the remaining service life of the cable is evaluated according to formula (16). When the accuracy correction term α is 0.2, all predicted values ​​fall between the upper and lower boundaries, and the average absolute error of the prediction result is 0.15 years. The warning threshold of the remaining service life is set to 4 years. When the predicted remaining service life of the cable is less than 4 years, an early fault warning is issued.

[0103] (1-α)·L a ≤L p ≤(1+α)·L a (16)

[0104] In the formula, α is the accuracy correction term, L a is the actual value of the remaining service life of the cable, L p It is the predicted value of the remaining service life of the cable.

[0105] In a specific implementation, a cable early fault warning system based on multi-feature fusion and iterative prediction model includes:

[0106] Standardization processing module: used to standardize the time series data of early fault characteristics and filter out important features;

[0107] Fusion module: used to calculate the weights of multiple important features and fuse multiple important features into comprehensive features;

[0108] Model building module: used to combine the chaotic grey wolf optimization-cuckoo search hybrid algorithm with the support vector regression model to build an augmented support vector regression model;

[0109] Iterative prediction module: used to train the augmented support vector regression model using comprehensive features, and iteratively predict the test data based on the trained augmented support vector regression model;

[0110] Fault warning module: used to provide early fault warning through the comprehensive feature threshold and iterative prediction results corresponding to cable failure.

[0111] In a specific embodiment, the fusion module further includes:

[0112] The feature extraction unit is used to extract early fault features of cables from multi-source data, including electrical features, thermal features, optical features, etc.

[0113] The weight calculation unit is used to calculate the weight of each feature by using a subjective and objective comprehensive weight method according to the time series of the known cable early fault features.

[0114] The feature fusion unit is used to fuse the weighted early fault features to generate comprehensive features.

[0115] In a specific embodiment, the iterative prediction module further includes:

[0116] The model training unit is used to train the augmented support vector regression model using the time series of the comprehensive features of the known early cable faults.

[0117] The iterative prediction unit is used to iteratively predict the change trend of the unknown comprehensive characteristics to update the prediction result of the remaining service life of the cable.

[0118] In a specific embodiment, the fault warning module further includes:

[0119] The failure threshold determination unit is used to determine the comprehensive characteristic threshold corresponding to the cable failure according to the type and operating conditions of the cable.

[0120] The life calculation unit is used to calculate the remaining service life of the cable according to the iterative prediction results of the comprehensive characteristics and the failure threshold.

[0121] In a specific embodiment, it also includes a visual warning module, which is used to visually display the comprehensive feature iterative prediction curve, the cable remaining service life prediction result and the early fault warning status.

[0122] In a specific embodiment, the visual warning module further includes:

[0123] The result display unit is used to display the comprehensive feature iterative prediction curve, the cable remaining service life prediction result and the early fault warning status in the form of charts, curves, etc.

[0124] The user interaction unit allows the user to interact with the system through the interface, such as adjusting the prediction model parameters, viewing the early fault warning results of different cables, etc.

[0125] Through the collaborative work of the above modules, the cable early fault warning system based on multi-feature fusion and iterative prediction model of the present invention can achieve reliable and accurate early warning of cable faults, providing a scientific basis for cable maintenance and management.

[0126] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0127] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A cable early fault warning method based on multi-feature fusion and iterative prediction model, characterized in that: include: Standardize the time series data of early fault characteristics and filter out important features; Calculate the weights of multiple important features and merge them into comprehensive features; Combining the Chaos Grey Wolf Optimization-Cuckoo Search Hybrid Algorithm with the Support Vector Regression Model, an Augmented Support Vector Regression Model is constructed. The augmented support vector regression model is trained using comprehensive features, and the data to be tested is iteratively predicted based on the trained augmented support vector regression model; Early fault warning is carried out through the comprehensive feature threshold corresponding to cable failure and the iterative prediction results.

2. According to claim 1, a cable early fault warning method based on multi-feature fusion and iterative prediction model is characterized in that: The performing of the standardization process to screen out the important features includes: respectively calculating the early fault features with a decreasing trend and the early fault features with an increasing trend.

3. A cable early fault warning method based on multi-feature fusion and iterative prediction model according to claim 2, characterized in that: The standardization process comprises: In the formula, H d is an early fault feature with a decreasing trend, H i is an early fault feature with an increasing trend, H t is the early fault characteristic value at time t, H0 is the initial early fault characteristic, H crit It is the key value of early fault characteristics.

4. The cable early fault warning method based on multi-feature fusion and iterative prediction model according to claim 1 is characterized in that: The step of calculating the weights of multiple important features and fusing the multiple important features into a comprehensive feature includes: calculating the weights of multiple important features by a subjective and objective comprehensive weight method, wherein the subjective and objective comprehensive weight method is obtained by combining the fuzzy hierarchical analysis method and the entropy weight method, obtaining a comprehensive weight based on the weights of multiple important features, and calculating the comprehensive feature based on the comprehensive weight.

5. A cable early fault warning method based on multi-feature fusion and iterative prediction model according to claim 4, characterized in that: The fusion into comprehensive features includes: In the formula, α k is the weight corresponding to the early fault feature calculated based on the fuzzy analytic hierarchy process, β k is the weight corresponding to the early fault feature calculated based on the entropy weight method, f ij is the element in the fuzzy consistent matrix, n is the number of early fault features, E k is the entropy value of early fault characteristics, w k is the comprehensive weight corresponding to the kth early fault feature; H c =H1w1+H2w2+…+H k w k ; In the formula, H c is a comprehensive feature, H k is the kth early fault feature.

6. The cable early fault warning method based on multi-feature fusion and iterative prediction model according to claim 1 is characterized in that: The chaotic gray wolf optimization-cuckoo search hybrid algorithm includes: updating the initial position of the gray wolf individual; updating the position of the gray wolf individual again through the bird nest position in the cuckoo search algorithm; and determining the updated position of the common gray wolf according to the optimal solution of updating the position of the gray wolf individual again.

7. A cable early fault warning method based on multi-feature fusion and iterative prediction model according to claim 6, characterized in that: The step of determining the updated position of the common gray wolf according to the optimal solution for re-updating the position of the individual gray wolf comprises: Update the initial positions of the gray wolf individuals (α, β, and δ): X1=X α (t)-A1|C1X α (t)-mX(t)|; X2=X β (t)-A2|C2X β (t)-mX(t)|; X3=X δ (t)-A3|C3X δ (t)-mX(t)|; Among them, X α , X β , X δ are the position vectors of α wolf, β wolf and δ wolf respectively, m is the chaotic variable, A1, A2, A3, C1, C2 and C3 are the coefficient vectors, t is the current iteration number, X1, X2 and X3 are the first three optimal solutions; The position of the individual gray wolf is updated again through the updating formula of the bird nest position in the cuckoo search algorithm; Where i is the number of individuals in the current wolf pack, τ is the step length control parameter, S is the Lévy flight search step length, and x n For the location of the bird's nest; Finally, the updated position of the common gray wolf is obtained according to the formula; X(t+1)=(X1+X2+X3) / 3; Where X is the updated position of the common gray wolf.

8. The cable early fault warning method based on multi-feature fusion and iterative prediction model according to claim 1 is characterized in that: The construction of the augmented support vector regression model includes: S1: Set the initial division ratio, divide the acquired early fault feature time series into two sections, and obtain the front and back time series data sets; S2: Set the initial time scale, kernel scale parameters and box constraint parameters, and use the previous time series to train the support vector regression model; S3: Use the trained support vector regression model to iteratively predict the data of the next period of time, and use the root mean square error between the predicted value and the true value of the next period of time as the objective function, and use the chaotic gray wolf optimization-cuckoo search hybrid algorithm to determine l optimal parameters to be determined; S4: Substitute the l best unknown parameters into the support vector regression model to establish an augmented support vector regression model.

9. The cable early fault warning method based on multi-feature fusion and iterative prediction model according to claim 1 is characterized in that: The iterative prediction of the test data based on the trained augmented support vector regression model includes: when the time scale of the input variable in the iterative prediction process is p, it means that the previous p time series points are used to predict the next time series point, and the predicted value and its previous p-1 time series points are reorganized into a new time series group to predict the next moment; iteratively update p time series until the prediction of the qth time series point is completed.

10. A cable early fault warning system based on multi-feature fusion and iterative prediction model, characterized in that: include: Standardization processing module: used to standardize the time series data of early fault characteristics and filter out important features; Fusion module: used to calculate the weights of multiple important features and fuse multiple important features into comprehensive features; Model building module: used to combine the chaotic grey wolf optimization-cuckoo search hybrid algorithm with the support vector regression model to build an augmented support vector regression model; Iterative prediction module: used to train the augmented support vector regression model using comprehensive features, and iteratively predict the test data based on the trained augmented support vector regression model; Fault warning module: used to provide early fault warning through the comprehensive feature threshold and iterative prediction results corresponding to cable failure.

Images