SOM-based power cable multi-parameter aging evaluation method and system

Abstract

This invention discloses a multi-parameter aging assessment method and system for power cables based on SOM (Self-Organizing Mechanism). The method includes: acquiring cable samples in different aging states; extracting peel force, isothermal relaxation aging factor, initial discharge voltage, and breakdown voltage at different holding times from unaged cable samples and cable samples in different aging states; training a SOM modified with PSO (Proof-of-Synthesis) to obtain a multi-parameter cable aging state assessment model, wherein PSO uses a cosine law decreasing strategy for weight update; and inputting the characteristic parameters of the cable to be assessed into the assessment model to obtain the aging state assessment result. This invention introduces mechanical and electrical multi-dimensional parameters as aging factors to judge the insulation aging state. The PSO search process uses a cosine law decreasing strategy for weight search, which improves efficiency and classification accuracy, accurately reflects the aging state assessment of cables under different production levels and operating environments, and has better universality.

Description

Technical Field

[0001] This invention belongs to the field of cable aging assessment technology, and relates to a multi-parameter aging assessment method and system for power cables based on SOM (Self-Organizing Map Neural Network). Background Technology

[0002] Due to the requirements of space corridors and environmental protection, power cables, with their excellent electrical and mechanical properties, have been widely used in long-distance power transmission and offshore power transmission, gradually becoming an important carrier and key equipment for power transmission in power systems. During actual operation, factors such as ambient temperature, external forces, and internal defects in the cable can affect its insulation performance, leading to a weakening of its electrical properties and thus reducing its service life. Therefore, regular aging condition assessments of cables are an important means of ensuring the safe and reliable operation of power systems.

[0003] The aging state of cables is typically related to their breakdown voltage. Related aging state assessment schemes involve destructive measurements. For example, patent CN115032488A uses environmental data and characteristic breakdown times of cable samples to establish a multi-physics-based prediction model for the aging life of high-voltage submarine cable insulation. However, this model is only applicable to the cables tested and lacks universality. Non-destructive aging state assessment schemes, such as patent CN111610407B, use parameters such as cable operating time, partial discharge, and dielectric loss to assess cable aging state. However, these parameters are significantly affected by the cable's operating environment and cannot be directly linked to physical quantities directly related to cable life. Furthermore, this assessment strategy requires known cable aging state, is supervised learning, and is not suitable for judging the aging state of unknown cable types. Patent CN114184903B uses electrical treeing factors to assess cable aging state, requiring cable slicing and cannot be directly applied to operating cables.

[0004] In addition, isothermal relaxation current can also achieve non-destructive aging condition assessment of cables. For example, patent CN110231511B obtains the aging factor by fitting the isothermal relaxation current method; patent CN113777138B improves this scheme. This measurement scheme has corresponding standards, and the applied voltage is lower than the operating voltage, causing no damage to the cable. However, current cable aging condition assessment methods based on isothermal relaxation current only use the aging factor as the assessment standard, resulting in a single evaluation criterion that is greatly affected by the cable's manufacturing level and operating environment. On the other hand, the obtained aging factor cannot be matched with conventional cable testing parameters, such as the cable life index characterized by breakdown voltage. Furthermore, in traditional isothermal relaxation aging assessment, the aging factor is obtained by fitting and calculating the current data through the depolarization current obtained from the test, and then judging the aging state of the cable. This is greatly affected by the cable's manufacturing level and operating environment, resulting in an inaccurate assessment of the insulation aging state. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a multi-parameter aging assessment method and system for power cables based on SOM (Self-Organizational Mechanism). It introduces peelability test, isothermal relaxation current, partial discharge initiation voltage, and step-by-step withstand voltage test, extracts aging factors such as peel force and isothermal relaxation current, initiation discharge voltage, and breakdown voltage as aging characteristic parameters, and constructs a multi-parameter cable aging state assessment model through SOM, thereby improving the accuracy and universality of the assessment.

[0006] The present invention adopts the following technical solution.

[0007] A multi-parameter aging assessment method for power cables based on SOM (Self-Analysis and Oxidation Method) includes the following steps:

[0008] Step 1: Aging cable samples from different manufacturers under different environments to obtain aged cable samples with different manufacturing levels, operating environments, and operating times;

[0009] Step 2: Extract mechanical and electrical multidimensional characteristic parameters from unaged cable samples and aged cable samples under different manufacturing levels, operating environments and operating times, including peel force, isothermal relaxation aging factor, initial discharge voltage and breakdown voltage under different holding times.

[0010] Step 3: Train the SOM improved by PSO using the extracted feature parameters to obtain a multi-parameter cable aging condition assessment model, in which PSO uses a cosine law decreasing strategy to update the weights.

[0011] Step 4: Extract the characteristic parameters of the cable to be evaluated, input them into the multi-parameter cable aging state evaluation model, and obtain the aging state evaluation results.

[0012] Preferably, in step 1, several cable samples produced by different manufacturers are randomly divided into 5 groups, one of which is not aged, and the other 4 groups are aged in air for 6 months, in air for 12 months, in water for 6 months, and in water for 12 months, respectively, to obtain aged cable samples under different manufacturing levels, operating environments and operating times.

[0013] Preferably, the voltage applied during aging in air and water is an alternating voltage, and the applied voltage is a voltage that causes the temperature of the cable conductor in the air to reach 90°C.

[0014] Preferably, in step 2, the sample is subjected to a peelability test of its insulation shield to extract the peel force, as follows:

[0015] The mechanical stress, denoted as peel force F, is measured at room temperature using a tensile testing machine to separate the insulating shield strip of the sample from the insulating layer by at least 100 mm, under conditions of a peel angle of 180° and a speed of 250 mm / min. N .

[0016] Preferably, in step 2, the sample is subjected to isothermal relaxation current testing, and the aging factor is extracted from the isothermal relaxation current, as follows:

[0017] 1) At room temperature, the sample was polarized with a polarization voltage of 1 kV and a voltage application time of 1800 s;

[0018] 2) After polarization, the sample is short-circuited for 1800s. The current during the short-circuit process is collected and stored to obtain the isothermal relaxation current I(t).

[0019] 3) The isothermal relaxation current I(t) is fitted with the following third-order exponential decay function to obtain the optimal fitting parameters:

[0020]

[0021] In the formula, I0 is the steady-state current, a i and τ i The fitting parameters are i = 1, 2, 3; t is time; the empirical ranges of the fitting parameters τ1, τ2, and τ3 are set to 0–20s, 10–100s, and 50–500s, respectively.

[0022] 4) Calculate the aging factor based on the optimal fitting parameters.

[0023]

[0024]

[0025] Where Q2(τ2) is the determining factor of the interface polarization between the crystalline region and the amorphous region, and Q3(τ3) is the determining factor of the polarization caused by aging.

[0026] Preferably, in step 2, the partial discharge initiation voltage of the sample is detected to obtain the initiation discharge voltage U. PDIV The details are as follows:

[0027] The partial discharge initiation voltage was tested using the pulsed current method. The sample voltage was linearly increased at a rate of 80 V / s until the partial discharge quantity exceeded 10 pV. The voltage at this point was recorded as the initiation discharge voltage U. PDIV .

[0028] Preferably, in step 2, the sample is subjected to a withstand voltage test with progressively increasing voltage, and the breakdown voltage U at holding times of 1 min and 10 min is extracted. bd(1min) U bd(10min) The details are as follows:

[0029] Using 8.7 kV as the starting voltage, the sample was voltaged at a rate of 1 kV / s with 1 kV increments. After reaching the target voltage, the voltage was held for 1 minute before continuing the voltage increase until the sample broke down. The breakdown voltage at the 1-minute holding time was recorded as U. bd(1min) The initial voltage, boost rate, and boost step remain constant. After each boost to the target voltage, the voltage is held for 10 minutes before the voltage is increased again until the sample breaks down. The breakdown voltage at the 10-minute holding time is recorded as U. bd(10min) .

[0030] Preferably, in step 3, the extracted feature parameter data is normalized to form a sample set. The sample set is then trained using a PSO-modified SOM to obtain a multi-parameter cable aging condition assessment model. The process of using PSO to modify the SOM is as follows:

[0031] Step (1): Initialize and set the network structure of the SOM;

[0032] Step (2): Initialize the particle swarm weights, learning factor, maximum allowed number of iterations, initial position and initial velocity of each particle;

[0033] Step (3): Update the velocity and position of all particles in the particle swarm, and update the weight ω using a cosine law decreasing strategy;

[0034] Step (4): Calculate the fitness of each particle with corresponding weights, and update the global optimal position and its corresponding weights and the optimal position of each particle according to the fitness, so as to use for iterative updates of particle velocity and position;

[0035] Step (5): Determine whether the maximum allowed number of iterations or the expected fitness value requirement is met. If it is met, the iteration stops and the global optimal result is output, which is the optimal weight parameter after PSO optimization; otherwise, return to step (3).

[0036] Step (6): Use the optimal weight parameters after PSO optimization as the weight parameters of the SOM network to construct the SOM network for evaluation and analysis of cable aging status.

[0037] Preferably, the formula for updating the weight ω using the cosine law decreasing strategy is as follows;

[0038] ω=ω min +(ω max -ω min ){[1+cos((i ter -1)π / (i max -1))] / 2}

[0039] Where, ω max ω is the maximum weight at the start of the search. min The minimum weight at the end of the search; i ter The number of steps performed in the iteration; i max This represents the maximum allowed number of iterations.

[0040] A multi-parameter aging assessment system for power cables based on SOM (System Oscillator) includes:

[0041] The cable sample acquisition module is used to age cable samples produced by different manufacturers in different environments to obtain aged cable samples under different manufacturing levels, operating environments and operating times.

[0042] The feature parameter extraction module is used to extract mechanical and electrical multidimensional feature parameters from unaged cable samples and aged cable samples under different manufacturing levels, operating environments and operating times, including peel force, isothermal relaxation aging factor, initiation discharge voltage and breakdown voltage under different holding times.

[0043] The model training module is used to train the SOM improved by PSO using the extracted feature parameters to obtain a multi-parameter cable aging state assessment model, in which PSO uses a cosine law decreasing strategy to update the weights.

[0044] The aging condition assessment module is used to extract the characteristic parameters of the cable to be assessed, input the multi-parameter cable aging condition assessment model, and obtain the aging condition assessment results.

[0045] A terminal includes a processor and a storage medium; the storage medium is used to store instructions.

[0046] The processor is configured to operate according to the instructions to execute the steps of the method.

[0047] A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method.

[0048] The beneficial effects of this invention are compared with those of the prior art:

[0049] 1. This invention ages cables manufactured by different domestic and foreign companies under different environments to obtain cable samples in different aging states. The samples before and after aging are then subjected to peelability tests, isothermal relaxation current tests, partial discharge initiation voltage tests, and progressive withstand voltage tests. The mechanical and electrical multi-dimensional characteristics, such as peel force, isothermal relaxation current aging factor, initiation discharge voltage, and breakdown voltage, are extracted as comprehensive parameters reflecting the cable's aging state. This invention can accurately reflect the aging state assessment of cables under different production levels and operating environments. It is applicable to the aging assessment of cables under different manufacturing levels and operating environments (water and air), and has a wide range of applications. By introducing different mechanical (peel force) and electrical (initiation discharge, withstand voltage breakdown, etc.) multi-dimensional parameters as aging factors to judge the insulation aging state, it overcomes the problem of inaccurate aging assessment caused by traditional methods that rely solely on a single isothermal relaxation aging factor.

[0050] 2. This invention uses the extracted feature parameters as input parameters, constructs the relationship between different parameters through SOM, and uses PSO method to search parameters in the SOM network construction process. In the PSO search process, a cosine law decreasing strategy is used to search the weights. The weights can maintain a large value for a long time in the early stage of the search to improve search efficiency, and maintain a small value for a long time in the later stage of the search to improve search accuracy. This effectively improves efficiency and classification accuracy, and has better universality for cable aging status assessment. Attached Figure Description

[0051] Figure 1 This is an overall flowchart of the evaluation method of the present invention;

[0052] Figure 2 Overall cable wiring diagram for isothermal relaxation current testing;

[0053] Figure 3 Wiring diagram for testing the partial discharge initiation voltage using the pulse current method;

[0054] Figure 4 Flowchart for SOM implementation;

[0055] Figure 5 The image shows the training results of cable aging condition assessment based on SOM.

[0056] Figure 6The figure shows the test results of cable aging condition assessment based on SOM. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0058] like Figure 1 As shown in the figure, this embodiment involves a multi-parameter aging assessment method for power cables based on SOM (Self-Analysis of Aging). The specific steps are as follows:

[0059] Step 1: Aging cable samples from different manufacturers under different environments to simulate cable samples with different aging degrees under different manufacturing levels, operating environments and operating times;

[0060] Several cable samples from different manufacturers were randomly divided into 5 groups. One group was not aged, while the other 4 groups were aged in air for 6 months, in air for 12 months, in water for 6 months, and in water for 12 months, respectively, to obtain aged cable samples under different manufacturing levels, operating environments, and operating times.

[0061] In this embodiment, to ensure the method's universal applicability, 14 8.7 / 10kV power cables manufactured by different domestic and international companies were selected and numbered A to N according to their manufacturers. The cables were aged in air and water for 6 months and 12 months respectively, resulting in five categories of samples under different aging times and environments. These were denoted by numbers 1 to 5, representing the aging type as: no aging, 6 months of air aging, 12 months of air aging, 6 months of water aging, and 12 months of water aging.

[0062] The voltage applied during aging is an alternating current voltage. The voltage is applied to raise the temperature of the cable conductor in the air to 90°C, and the same voltage is applied to the cable sample in the water.

[0063] Step 2: Extract multi-dimensional mechanical and electrical characterization parameters from unaged cable samples and aged cable samples under different manufacturing levels, operating environments and operating times, including peel force, isothermal relaxation aging factor, initiation discharge voltage and breakdown voltage.

[0064] This step involves conducting peelability tests, isothermal relaxation current tests, partial discharge initiation voltage tests, and mechanical and electrical tests on cable samples under different aging conditions (samples before and after aging). Peel force, isothermal relaxation aging factor, initiation discharge voltage, and breakdown voltage are extracted as characteristic parameters; details are as follows:

[0065] (1): The insulation shielding peelability test was performed on the sample, and the peel force characteristic value was extracted.

[0066] The peel force is the result of the peelability test of the insulation shield according to Clause 19 of GB / T 12706.2-2008. The test, conducted at room temperature using a tensile testing machine, separates the strip of the insulation shield from the insulation layer by at least 100 mm. The mechanical stress under these conditions—a peel angle of approximately 180° and a speed of 250 mm / min—is denoted as the peel force F. N Specifically, this includes:

[0067] 1) The test was conducted at room temperature. Three points approximately 120° apart along the circumference of the same cable sample were tested. The average tensile force obtained from the three tests was taken as the characteristic value F of the peel force of the sample. N .

[0068] 2) Make two parallel cuts 10mm apart on the surface of the extruded insulation shield of the sample in the longitudinal direction of the insulation, and pull out a strip 50mm long and 10mm wide along the direction parallel to the insulation core.

[0069] 3) Vertically mount the insulated wire core onto the tensile testing machine, clamp one end of the insulated wire core with one clamp, and clamp the 10mm strip with the other clamp;

[0070] 4) Apply a tensile force through the clamp to separate the 10mm strip from the insulation, pulling it apart by at least 100mm. Measure the tensile force F under conditions of an approximately 180° peel angle and a speed of 250mm / min. N Record the tension F N As a characteristic value of peeling force.

[0071] (2): The sample is subjected to isothermal relaxation current testing, and the characteristic value A of the aging factor is extracted based on the isothermal relaxation current. The testing system is as follows: Figure 2 As shown, Figure 2 In this method, the depolarization current acquisition device is directly connected to the sample, so that the depolarization current released by the cable insulation during depolarization is collected by the ammeter, while the depolarization current of the high-voltage insulated wire does not flow through the ammeter, thus making the collected depolarization current more accurate. Specifically:

[0072] 1) When switch 1 is placed in DC high voltage source circuit 2, the cable 3, i.e. the sample, is polarized. The polarization voltage is 1kV and the voltage application time is 1800s.

[0073] 2) After polarization, switch 1 to the grounding side 4 to short-circuit cable 3 for 1800s. The current during the short circuit is collected and stored by the current acquisition system 5. The collected current is the isothermal relaxation current I(t), and the current result during the short circuit is taken as the isothermal relaxation current I(t).

[0074] The above tests were conducted at room temperature of 298K, with a sampling rate of 50 sampling points / second.

[0075] To avoid interference during on-site testing, the testing system is shielded from on-site interference by a metal casing 6 and transmitted via a high-voltage coaxial cable 7.

[0076] 3) Extracting the aging factor characteristic value A from the isothermal relaxation current.

[0077] The isothermal relaxation current I(t) obtained from the test was fitted with a third-order exponential decay function. Where I0 is the steady-state current, a i and τ i The fitting coefficients are related to the three types of polarization in the medium. a1 and τ1 represent the bulk polarization components of the test sample during the test, a2 and τ2 represent the polarization components between the crystalline and amorphous regions in the test sample, and a3 and τ3 represent the polarization components caused by aging of the test sample.

[0078] The empirical ranges for the first, second, and third time constants v1, v2, and τ3 are set to 0–20 s, 10–100 s, and 50–500 s, respectively. A triple exponential decay model is then used to determine the goodness of fit R. 2 The optimal fitting parameters are determined with a target value close to 1. Here, goodness of fit refers to how well the regression line fits the observed values. 2 = 1 - "the ratio of the regression sum of squares to the total sum of squares". The closer the goodness of fit is to 1, the better the fit.

[0079] Based on the fitting results, using Calculate aging factors;

[0080] Among them, the determining factor of the interface polarization between the crystalline region and the amorphous region. Determinant of polarization caused by aging

[0081] The larger the aging factor A is, the more severe the aging state of the material. The aging factor A is recorded as a characteristic value.

[0082] (3): The partial discharge initiation voltage of the sample was detected.

[0083] like Figure 3 As shown, the pulsed current method is used for partial discharge signal detection. Specifically, the voltage is linearly increased at a rate of 80V / s until the partial discharge quantity exceeds 10p. The voltage at this point is recorded as the partial discharge initiation voltage, i.e., the initiation discharge voltage U. PDIV .

[0084] The MPD600 pulse current method partial discharge analyzer has a measurement frequency band of 10kHz to 1MHz, a measurement range of 0.1pC to 10000nC, and a measurement sensitivity of 0.1pC.

[0085] (4): The sample was subjected to a withstand voltage test with progressively increasing voltage, and the voltage at which the cable broke down was extracted at holding times of 1 min and 10 min.

[0086] The experiment started with a voltage of 8.7 kV (set according to the cable's voltage rating). The voltage was increased by 1 kV at a rate of 1 kV / s each time. After reaching the target voltage each time, the voltage was held for 1 minute before continuing to increase. The experiment ended with breakdown. The breakdown voltage U was recorded after a holding time of 1 minute. bd(1min) Similarly, under the same starting voltage, boost rate, and boost step, the breakdown voltage after each boost with a hold time of 10 minutes is denoted as U. bd(10min) .

[0087] Step 3: Train the SOM optimized by PSO using the extracted feature parameters to obtain a multi-parameter cable aging condition assessment model, where PSO uses a cosine law decreasing strategy to update the weights.

[0088] This step first normalizes and preprocesses the datasets formed by the feature parameters of tests at different aging levels to obtain a sample set. Then, an adaptive network is constructed, using the sample set as input parameters. A multi-parameter cable aging state assessment model is built by training the SOM (Structured Object Model) to evaluate the aging state of the cable. Since the SOM network is highly dependent on the initial weight settings, it affects the network's convergence speed and learning effect. If the weights of a neuron differ significantly from the input vector pattern, it may never become the winning neuron, and its weights cannot be effectively trained, thus reducing the convergence speed and clustering accuracy of the network. The PSO (Programmable Optimization Sort) algorithm is used to optimize the neural network. The weight vector of the SOM neural network is treated as individual particles, and the weight vector is updated by updating the position of each particle, using a cosine law decreasing strategy for weight updates. After a certain number of iterations, the optimized SOM network weight vector is used to find the winning neuron, forming the SOM network.

[0089] (1) Normalize the aging characteristic data.

[0090] The dataset contains 100 sets of data from cable samples in unaged and different aging states, with each set including 5 characteristic parameters: peel force F. N Isothermal relaxation aging factor A, initiation discharge voltage U PDIV Breakdown voltage U with a holding time of 1 minute bd(1min) and the breakdown voltage U with a holding time of 10 minutes bd(10min) That is, the five feature parameters of the i-th data set are represented as x. ij j = 1, 2, ..., 5, after normalization, construct matrix X = (x ij ) 100×5 , i = 1, 2, ..., 100.

[0091] (2) Train the PSO-modified SOM network, the specific steps are as follows: Figure 4 As shown.

[0092] The SOM algorithm described above simulates the self-organizing feature mapping function of the brain's neural system. It is a feedforward network employing unsupervised competitive learning. Through learning, it can extract important features or inherent patterns from a set of data, classifying them in a discrete-time manner. It can map arbitrarily high-dimensional inputs to a low-dimensional space, and make certain similarities within the input data manifest as geometrically adjacent feature maps. However, the SOM network is highly dependent on the initial weight settings, which can affect the network's convergence speed and learning performance. Therefore, this invention uses the PSO algorithm to adjust the weights for better classification results.

[0093] The training process of the PSO-improved SOM is as follows:

[0094] Suppose there exists a population X = (x1, x2, ..., xn) consisting of n particles in D-dimensional space. n ), where the velocity of the i-th particle is represented by V. i =(v i1 ,v i2 ,...,v iD ) T This indicates that the individual extreme value and the population extreme value of the particle are P, respectively. i =(p i1 ,p i2 ,...,p iD ) T and P g =(p g1 p g2 ,...,p gD ) T .

[0095] Step (1): Initialize the structure of the SOM network. Since each set of data has 5 feature parameters, set the number of nodes in the network input layer to 5.

[0096] Select the output layer based on empirical formulas. Where i train Let i be the number of training samples train =100, which defines a 7×7 two-dimensional self-organizing map network.

[0097] Step (2): Initialize and set the particle swarm weights, learning factor, maximum allowed number of iterations, initial position and initial velocity of each particle, etc.

[0098] Step (3): Update the velocity and position of all particles in the particle swarm:

[0099]

[0100]

[0101] Where ω is the inertia weight; k is the current iteration number; c1 and c2 are non-negative constants, which are learning factors; r1 and r2 are random numbers uniformly distributed on [0, 1].

[0102] Since the weight ω enables particles to maintain their inertia, it plays a crucial role in the convergence of the algorithm. Larger weights result in faster convergence but are less likely to yield an accurate solution; smaller weights are beneficial for local searches but result in slower convergence. Traditional PSO particle search uses a linearly decreasing weight update method, but this method results in uniform weight changes, which is detrimental to improving search accuracy and cannot meet the requirement of fast initial search speed followed by slower later search speeds. This invention utilizes a cosine-law decreasing strategy for weight ω updates. The weights maintain a larger value for a longer period in the early stages of the search to improve search efficiency, and a smaller value for a longer period in the later stages to improve search accuracy. The specific weight ω update formula is as follows:

[0103] ω=ω min +(ω max -ω min ){[1+cos((i ter -1)π / (i max -1))] / 2}

[0104] Where, ω max ω is the largest ω at the start of the search; min The minimum ω at the end of the search; i ter The number of steps performed in the iteration; i max This represents the maximum allowed number of iterations.

[0105] Step (4): Calculate the fitness of each particle with corresponding weights, and update the global optimum, the optimum of each particle and its corresponding weights.

[0106] Fitness was obtained using Euclidean distance;

[0107] The local optimum (the optimal position of each particle) is determined and updated based on fitness. id and the global optimal value p gd And the corresponding weight ω, used for subsequent iterative updates of particle velocity and position.

[0108] Step (5): Determine the number of iterations or the expected fitness value requirement. If the stopping condition is met, the search stops and the global optimal result is output, which is the optimal weight parameter after PSO optimization. Otherwise, return to step (3) to continue the search iteration.

[0109] Step (6): Use the optimal weight parameters after PSO optimization as the weight parameters of the SOM network, construct the SOM network, and perform cluster analysis.

[0110] In this embodiment, the target error value is set to 0.05, the initial learning rate is 0.5, the initial neighborhood range is 6, and the final neighborhood value is 1. The maximum and minimum values ​​of the inertia weight in the particle swarm optimization parameters are set to 0.9 and 0.2, respectively, the learning factor is c1 = c2 = 0.8, and the number of iterations is 200.

[0111] After the training process of the self-organizing network, the specific relationship between each neuron in the input layer and each input pattern class is completely determined. Therefore, it can be used as a classifier for assessing the aging status of cables. The aging status is divided into three categories according to its degree: non-aging, moderate aging, and severe aging.

[0112] The 2D SOM output node diagram shows the assessment results of the aging state. The 2D SOM output nodes are as follows: Figure 5 As shown, the training samples are divided into three categories: unaged, moderately aged, and severely aged, with 22, 51, and 27 samples respectively.

[0113] To verify the reliability of the SOM-based multi-parameter aging condition assessment method for power cables, a test set of 15 field-operated samples was selected for testing. This set included 7 unaged samples, 5 moderately aged samples, and 3 severely aged samples. The training results were obtained using the test set data, as shown below. Figure 6 As shown, the results indicate that the aging status is accurately determined.

[0114] Step 4: Extract the characteristic parameters of the cable to be evaluated, input them into the multi-parameter cable aging state evaluation model, and obtain the aging state evaluation results.

[0115] The working process of the trained SOM (Multi-parameter Cable Aging Status Assessment Model) is as follows: When any input pattern class that does not belong to the network during training is input into the network, the self-organizing network will be assigned to the closest pattern class to complete the clustering and partitioning. Then, the aging degree of each region is distinguished by human experience and sample values ​​of each region, thereby realizing the assessment of the aging status of the cable with input features.

[0116] Embodiment 2 of the present invention provides a multi-parameter aging assessment system for power cables based on SOM, comprising:

[0117] The cable sample acquisition module is used to age cable samples produced by different manufacturers in different environments to obtain aged cable samples under different manufacturing levels, operating environments and operating times.

[0118] The feature parameter extraction module is used to extract mechanical and electrical multidimensional feature parameters from unaged cable samples and aged cable samples under different manufacturing levels, operating environments and operating times, including peel force, isothermal relaxation aging factor, initiation discharge voltage and breakdown voltage under different holding times.

[0119] The model training module is used to train the SOM improved by PSO using the extracted feature parameters to obtain a multi-parameter cable aging state assessment model, in which PSO uses a cosine law decreasing strategy to update the weights.

[0120] The aging condition assessment module is used to extract the characteristic parameters of the cable to be assessed, input the multi-parameter cable aging condition assessment model, and obtain the aging condition assessment results.

[0121] A terminal includes a processor and a storage medium; the storage medium is used to store instructions.

[0122] The processor is configured to operate according to the instructions to execute the steps of the method.

[0123] A computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the method.

[0124] The beneficial effects of this invention are compared with those of the prior art:

[0125] 1. This invention ages cables manufactured by different domestic and foreign companies under different environments to obtain cable samples in different aging states. The samples before and after aging are then subjected to peelability tests, isothermal relaxation current tests, partial discharge initiation voltage tests, and progressive withstand voltage tests. The mechanical and electrical multi-dimensional characteristics, such as peel force, isothermal relaxation current aging factor, initiation discharge voltage, and breakdown voltage, are extracted as comprehensive parameters reflecting the cable's aging state. This invention can accurately reflect the aging state assessment of cables under different production levels and operating environments. It is applicable to the aging assessment of cables under different manufacturing levels and operating environments (water and air), and has a wide range of applications. By introducing different mechanical (peel force) and electrical (initiation discharge, withstand voltage breakdown, etc.) multi-dimensional parameters as aging factors to judge the insulation aging state, it overcomes the problem of inaccurate aging assessment caused by traditional methods that rely solely on a single isothermal relaxation aging factor.

[0126] 2. This invention uses the extracted feature parameters as input parameters, constructs the relationship between different parameters through SOM, and uses PSO method to search parameters in the SOM network construction process. In the PSO search process, a cosine law decreasing strategy is used to search the weights. The weights can maintain a large value for a long time in the early stage of the search to improve search efficiency, and maintain a small value for a long time in the later stage of the search to improve search accuracy. This effectively improves efficiency and classification accuracy, and has better universality for cable aging status assessment.

[0127] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0128] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0129] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0130] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A multi-parameter aging assessment method for power cables based on SOM, characterized in that: The method includes the following steps: Step 1: Aging cable samples from different manufacturers under different environments to obtain aged cable samples with different manufacturing levels, operating environments, and operating times; Step 2: Extract mechanical and electrical multidimensional characteristic parameters from unaged cable samples and aged cable samples under different manufacturing levels, operating environments and operating times, including peel force, isothermal relaxation aging factor, initial discharge voltage and breakdown voltage under different holding times. Step 3: Train the SOM improved by PSO using the extracted feature parameters to obtain a multi-parameter cable aging condition assessment model, in which PSO uses a cosine law decreasing strategy to update the weights. Using a cosine-law decreasing strategy for weighting The updated formula is: in, The largest at the start of the search ; Minimum at the end of the search ; The number of steps performed in the iteration; This represents the maximum allowed number of iterations. Step 4: Extract the characteristic parameters of the cable to be evaluated, input them into the multi-parameter cable aging status evaluation model, and obtain the aging status evaluation results.

2. The method for multi-parameter aging assessment of power cables based on SOM according to claim 1, characterized in that: In step 1, several cable samples from different manufacturers were randomly divided into 5 groups. One group was not aged, while the other 4 groups were aged in air for 6 months, in air for 12 months, in water for 6 months, and in water for 12 months, respectively, to obtain aged cable samples under different manufacturing levels, operating environments, and operating times.

3. The method for multi-parameter aging assessment of power cables based on SOM according to claim 2, characterized in that: The voltage applied during aging in air and water is an alternating current voltage, and the applied voltage is the voltage that causes the cable conductor temperature in the air to reach 90°C.

4. The method for multi-parameter aging assessment of power cables based on SOM according to claim 1, characterized in that: In step 2, the insulation shielding peelability test is performed on the sample, and the peel force is extracted, as follows: The mechanical stress, measured at room temperature using a tensile testing machine, at a peel angle of 180° and a speed of 250 mm / min, resulting in a strip of insulating shielding separating from the insulating layer of the sample at a distance of at least 100 mm, is denoted as the peel force. .

5. The method for multi-parameter aging assessment of power cables based on SOM according to claim 1, characterized in that: In step 2, the sample is subjected to isothermal relaxation current testing, and the aging factor is extracted from the isothermal relaxation current, as detailed below: 1) At room temperature, the sample was polarized with a polarization voltage of 1 kV and a voltage application time of 1800 s; 2) After polarization, the sample was short-circuited for 1800 s. The current during the short-circuit process was collected and stored to obtain the isothermal relaxation current. ; 3) For isothermal relaxation current The optimal fitting parameters are obtained by fitting a third-order exponential decay function as follows: In the formula, For steady-state current, and For the fitting parameters, i =1,2,3; For time; fitting parameters , , The experience ranges were set to 0~20s, 10~100s and 50~500s respectively; 4) Calculate the aging factor based on the optimal fitting parameters. ; in, It is the determining factor for the interface polarization between the crystalline and amorphous regions. The determining factor for polarization caused by aging.

6. The method for multi-parameter aging assessment of power cables based on SOM according to claim 1, characterized in that: In step 2, the partial discharge initiation voltage of the sample is detected to obtain the initiation discharge voltage U. PDIV The details are as follows: The partial discharge initiation voltage was tested using the pulsed current method. The sample voltage was linearly increased at a rate of 80 V / s until the partial discharge quantity exceeded 10 pV. The voltage at this point was recorded as the initiation discharge voltage U. PDIV .

7. The method for multi-parameter aging assessment of power cables based on SOM according to claim 1, characterized in that: In step 2, the sample is subjected to a withstand voltage test with progressively increasing voltage, and the breakdown voltage at which the sample breaks down is obtained at holding times of 1 min and 10 min. U bd(1min) , U bd(10min) The details are as follows: Using 8.7 kV as the starting voltage, the sample was voltaged at a rate of 1 kV / s with 1 kV increments. After reaching the target voltage, the voltage was held for 1 minute before continuing the voltage increase until the sample broke down. The breakdown voltage at a holding time of 1 minute was recorded. U bd(1min) The initial voltage, voltage ramp rate, and voltage ramp step remain constant. After each voltage ramp to the target voltage, the voltage is held for 10 minutes before the voltage ramp is continued until the sample breaks down. The breakdown voltage at the 10-minute holding time is recorded. U bd(10min) .

8. The method for multi-parameter aging assessment of power cables based on SOM according to claim 1, characterized in that: In step 3, the extracted feature parameter data are normalized to form a sample set. The sample set is then trained using a PSO-improved SOM to obtain a multi-parameter cable aging condition assessment model. The process of using PSO to improve the SOM is as follows: Step (1): Initialize and set up the network structure of the SOM; Step (2): Initialize and set the particle swarm weights, learning factor, maximum allowed number of iterations, initial position and initial velocity of each particle; Step (3): Update the velocity and position of all particles in the particle swarm, and apply a cosine law decreasing strategy to adjust the weights. renew; Step (4): Calculate the fitness of each particle with corresponding weights, and update the global optimal position and its corresponding weights and the optimal position of each particle according to the fitness, so as to use for iterative updates of particle velocity and position; Step (5): Determine whether the maximum allowed number of iterations or the expected fitness value requirement is met. If it is met, the iteration stops and the global optimal result is output, which is the optimal weight parameter after PSO optimization; otherwise, return to step (3). Step (6): Use the optimal weight parameters after PSO optimization as the weight parameters of the SOM network to construct the SOM network for evaluation and analysis of cable aging status.

9. A multi-parameter aging assessment system for power cables based on SOM, utilizing the method described in any one of claims 1-8, characterized in that: The system includes: The cable sample acquisition module is used to age cable samples produced by different manufacturers in different environments to obtain aged cable samples under different manufacturing levels, operating environments and operating times. The feature parameter extraction module is used to extract mechanical and electrical multidimensional feature parameters from unaged cable samples and aged cable samples under different manufacturing levels, operating environments and operating times, including peel force, isothermal relaxation aging factor, initiation discharge voltage and breakdown voltage under different holding times. The model training module is used to train the SOM improved by PSO using the extracted feature parameters to obtain a multi-parameter cable aging state assessment model, in which PSO uses a cosine law decreasing strategy to update the weights. The aging condition assessment module is used to extract the characteristic parameters of the cable to be assessed, input the multi-parameter cable aging condition assessment model, and obtain the aging condition assessment results.

10. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-8.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-8.

Images