A performance prediction method for ultra-high voltage cables

Abstract

The present invention relates to the technical field of power cables, and in particular to a performance prediction method for ultra-high voltage cables. The method includes steps such as data collection, data cleaning, constructing an algorithm model for optimizing support vector regression (SVR) using a genetic algorithm, and using the trained model to predict and evaluate performance. The method collects relevant data of ultra-high voltage cables, uses a density-based isolated point detection method to clean and normalize the data, optimizes the parameters and feature selection of SVR using a genetic algorithm, and finally predicts the performance of new cables and evaluates accuracy and reliability. The ultra-high voltage cable manufacturing process parameters include a variety of process parameters, and the performance data include electrical, mechanical, and thermal performance parameters. The present invention can effectively predict the performance of ultra-high voltage cables, provide a scientific basis for their design, manufacturing, operation and maintenance, and has important practical application value.

Description

Technical Field

[0001] The present invention relates to the technical field of power cables, and in particular to a performance prediction method for an ultra-high voltage cable. Background Art

[0002] With the continuous growth of electricity demand, ultra-high voltage cables play a vital role in power transmission. However, the performance of ultra-high voltage cables is affected by many factors, such as manufacturing process parameters, actual operating environment, etc., which makes it extremely challenging to accurately predict their performance. Existing performance prediction methods often have limitations and usually need to be carried out after the cable has been in operation for a period of time. This method is not only time-consuming and labor-intensive, but also difficult to accurately predict the performance of the cable in actual operation. In addition, the performance data of ultra-high voltage cables are usually diverse and complex, and may contain noise and outliers, which further increases the difficulty of accurate prediction. In order to ensure the stable operation and reliability of the power system, a more effective performance prediction method is needed to identify potential problems in advance and take corresponding measures. At present, although there are some related technologies, there is still room for improvement, such as how to process data more accurately and optimize models to improve prediction accuracy. Therefore, it is of great practical significance and application value to develop an ultra-high voltage cable performance prediction method based on advanced technology. Summary of the invention

[0003] In view of the technical problems existing in the prediction of the performance of ultra-high voltage cables mentioned above, the present invention proposes a method which is simple in method, easy in operation and can effectively and accurately predict the performance of ultra-high voltage cables.

[0004] In order to achieve the above object, the technical solution adopted by the present invention comprises the following steps:

[0005] S1. The first step is data collection, which involves collecting a large amount of data related to ultra-high voltage cables, including manufacturing process parameters and performance data in actual operation;

[0006] S2, then the density-based outlier detection method is used to clean the collected data and normalize the data;

[0007] S3. Then, according to the characteristics and requirements of the prediction of ultra-high voltage cable performance, an algorithm model for optimizing support vector regression (SVR) using genetic algorithm is constructed, and feature selection and parameter optimization are integrated. While optimizing SVR parameters, the selection of input features is considered;

[0008] S4. Finally, the trained model is used to predict the performance of the new ultra-high voltage cable. The relevant parameters of the cable are input, and the model will output the predicted performance indicators. The prediction results are evaluated and analyzed to determine the accuracy and reliability of the prediction.

[0009] The implementation steps of using genetic algorithm to optimize support vector regression SVR in step S3 are as follows:

[0010] S31. The first is the coding design. The chromosome coding of the individual is divided into two parts. One part is used to represent the parameters of SVR, the penalty coefficient C and the kernel function parameters, using real number coding; the other part is used for feature selection, using binary coding, each bit corresponds to a feature, 0 means not to select the feature, 1 means to select the feature;

[0011] S32, then randomly generate a certain number of individuals to form an initial population, each individual contains both the initial value of the SVR parameter and the initial state of feature selection;

[0012] S33, then define the fitness function, which takes into account both the prediction error and the number of selected features. Suppose the data set has n samples and the actual value is , the predicted value is , the number of selected features is m, the total number of features is N, then the fitness function is defined as , where MSE is the mean square error, is the weight coefficient used to balance the importance of prediction error and number of features;

[0013] S34, the next step is genetic operation. In the selection operation, the roulette wheel method is used to select excellent individuals to enter the next generation according to the fitness value of the individuals; in the crossover operation, arithmetic crossover is used for the SVR parameter part, and single-point crossover is used for the feature selection part; in the mutation operation, real number mutation is performed on the SVR parameter part, and binary mutation is performed on the feature selection part;

[0014] S35. Finally, steps S33 and S34 are repeated until a termination condition is met. The termination condition is that a preset number of iterations is reached, and finally the optimized SVR parameters and feature selection results are obtained.

[0015] Preferably, the ultra-high voltage cable manufacturing process parameters in step S1 include wire drawing process parameters, twisting process parameters, extrusion temperature, extrusion speed, extrusion pressure, wrapping process parameters, and welding process parameters of the metal shielding layer; the performance data include electrical properties, mechanical properties, and thermal performance parameters.

[0016] Preferably, the specific steps of using the density-based isolated point detection method to clean the collected data in step S2 are:

[0017] S21. First, for each data point , defining its k distance is the distance between the point and its kth nearest neighbor. Specifically, ,in yes The kth nearest neighbor of

[0018] S22, then calculate the reachable distance, for a data point and points in its neighborhood , the reachable distance is ,in Yes The k distance, Yes and The actual distance;

[0019] S23, then for the data point , whose local reachable density Defined as a point The reciprocal of the average of the reachable distances of the k nearest neighbors ,in Yes The number of k nearest neighbors of Yes Its neighboring points The reachable distance;

[0020] S24, then calculate the local outlier factor to measure the data point The difference between the density of and the density of other points in its neighborhood ,in Yes The local reachable density of

[0021] S25. Set a threshold value according to the value of the local outlier factor. are marked as outliers.

[0022] Preferably, the operation steps of using the trained model to predict and evaluate the performance of the new ultra-high voltage cable in step S4 are as follows:

[0023] S41. First, for the new ultra-high voltage cable data, its performance results are predicted through the model and its real performance index data is obtained through actual testing or measurement;

[0024] S42, then use the absolute error to calculate the error, and use the mean absolute error and the root mean square error as evaluation indicators to evaluate the accuracy of the prediction;

[0025] S43. Finally, the prediction results are evaluated and analyzed. According to the values ​​of the error index and the evaluation index, the prediction performance of the model is analyzed. If the error index exceeds the threshold of 8%, the model is further optimized by adjusting the model parameters, increasing the amount of data, and improving the feature selection. Different types of ultra-high voltage cables or different parameter ranges can also be analyzed separately to determine the performance of the model under different conditions.

[0026] Compared with the prior art, the advantages and positive effects of the present invention are that it can more effectively process noise and outliers in the performance data of ultra-high voltage cables, thereby improving data quality and reliability; by integrating feature selection with parameter optimization, the model can more accurately predict cable performance, thereby providing a more scientific basis for the design, manufacture, operation and maintenance of cables; by optimizing support vector regression SVR using a genetic algorithm, the generalization ability and prediction accuracy of the model are improved; different types of ultra-high voltage cables or different parameter ranges can be analyzed separately, thereby better adapting to various complex situations and ensuring the stable operation and reliability of the power system; the method has high flexibility and scalability, and can be adjusted and optimized according to actual needs, thereby providing strong support for the development of the field of power cable technology. DETAILED DESCRIPTION

[0027] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described below in conjunction with embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0028] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways than those described herein. Therefore, the present invention is not limited to the specific embodiments of the following disclosure.

[0029] Embodiment, with the continuous growth of electricity demand, ultra-high voltage cables are more and more widely used in power transmission systems. However, the performance of ultra-high voltage cables is affected by many factors, such as material quality, manufacturing process, operating environment, etc. In order to ensure the safe and stable operation of the power system, it is crucial to accurately predict the performance of ultra-high voltage cables. First, data collection is carried out to collect a large amount of relevant data of ultra-high voltage cables, including manufacturing process parameters and performance data in actual operation; the ultra-high voltage cable manufacturing process parameters include wire drawing process parameters, twisting process parameters, extrusion temperature, extrusion speed, extrusion pressure, wrapping process parameters, and welding process parameters of the metal shielding layer; performance data include electrical performance, mechanical performance, and thermal performance parameters.

[0030] In order to remove abnormal data from the collected data and improve the quality and reliability of the data, a density-based outlier detection method is used to clean the collected data. This method can identify abnormal points based on the density distribution of data points. Compared with other methods, it can more accurately detect abnormal data with large density differences from surrounding data points. First, for each data point , defining its k distance is the distance between the point and its kth nearest neighbor. Specifically, ,in yes The kth nearest neighbor of a data point; then calculate the reachable distance, and points in its neighborhood , the reachable distance is ,in Yes The k distance, Yes and The actual distance of the data point , whose local reachable density Defined as a point The reciprocal of the average of the reachable distances of the k nearest neighbors ,in Yes The number of k nearest neighbors of Yes Its neighboring points The reachable distance of the data point is then calculated to measure the local outlier factor The difference between the density of and the density of other points in its neighborhood ,in Yes The local reachable density; according to the value of the local outlier factor, set the threshold, when are marked as outliers.

[0031] In order to improve the accuracy of prediction of ultra-high voltage cable performance, and considering the selection of input features, a genetic algorithm is used to optimize the support vector regression SVR scheme. The first is the coding design, which divides the chromosome coding of the individual into two parts. One part is used to represent the parameters of SVR, the penalty coefficient C and the kernel function parameters, using real number coding; the other part is used for feature selection, using binary coding, each bit corresponds to a feature, 0 means that the feature is not selected, and 1 means that the feature is selected; secondly, a certain number of individuals are randomly generated to form an initial population, each individual contains both the initial value of the SVR parameter and the initial state of the feature selection; then the fitness function is defined, which takes into account the prediction error and the number of selected features at the same time. Assume that the data set has n samples and the actual value is , the predicted value is , the number of selected features is m, the total number of features is N, then the fitness function is defined as , where MSE is the mean square error, is the weight coefficient, which is used to balance the importance of prediction error and number of features. Next is the genetic operation. In the selection operation, the roulette wheel method is used to select excellent individuals to enter the next generation according to the fitness value of the individual. In the crossover operation, arithmetic crossover is used for the SVR parameter part, and single-point crossover is used for the feature selection part. In the mutation operation, the SVR parameter part is subjected to real number mutation, and the feature selection part is subjected to binary mutation. Finally, until the termination condition is met, the termination condition is to reach the preset number of iterations, and finally the optimized SVR parameters and feature selection results are obtained.

[0032] Finally, the trained model is used to predict and evaluate the performance of the new ultra-high voltage cable. First, for the new ultra-high voltage cable data, its performance results are predicted through the model and its real performance index data is obtained through actual testing or measurement; then the absolute error is used to calculate the error, and the mean absolute error and root mean square error are used as evaluation indicators to evaluate the accuracy of the prediction; finally, the prediction results are evaluated and analyzed, and the prediction performance of the model is analyzed according to the values ​​of the error index and the evaluation index. If the error is large, the model can be further optimized, such as adjusting the model parameters, increasing the amount of data, and improving feature selection. Different types of ultra-high voltage cables or different parameter ranges can also be analyzed separately to determine the performance of the model under different conditions.

[0033] Finally, when predicting and evaluating the performance of the new ultra-high voltage cable, the trained model needs to be used. Specifically, first, for the new ultra-high voltage cable data, use the model to predict its performance results, and obtain its real performance index data through actual testing or measurement; then use the absolute error to calculate the error, and use the mean absolute error and root mean square error as evaluation indicators to evaluate the accuracy of the prediction; finally, evaluate and analyze the prediction results, and analyze the prediction performance of the model based on the values ​​of the error index and the evaluation index. If the error is large, the model can be further optimized, such as adjusting the model parameters, increasing the amount of data, and improving feature selection. In addition, different types of ultra-high voltage cables or different parameter ranges can be analyzed separately to determine the performance of the model under different circumstances. Through the above examples, the implementation process of the ultra-high voltage cable performance prediction method is specifically demonstrated, which helps to verify the feasibility and effectiveness of the method and provide more reliable support for the application of ultra-high voltage cables in power systems.

[0034] The above description is only a preferred embodiment of the present invention and does not limit the present invention in other forms. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A performance prediction method for an ultra-high voltage cable, characterized in that: The following steps are involved: S1. First, data collection is performed to collect a large amount of relevant data of ultra-high voltage cables, including manufacturing process parameters and performance data in actual operation; the ultra-high voltage cable manufacturing process parameters include wire drawing process parameters, twisting process parameters, extrusion temperature, extrusion speed, extrusion pressure, wrapping process parameters, and welding process parameters of the metal shielding layer; the performance data include electrical performance, mechanical performance, and thermal performance parameters; S2, then the density-based outlier detection method is used to clean the collected data and normalize the data; S3. Then, according to the characteristics and requirements of the prediction of ultra-high voltage cable performance, an algorithm model for optimizing support vector regression (SVR) using genetic algorithm is constructed, and feature selection and parameter optimization are integrated. While optimizing SVR parameters, the selection of input features is considered; S4. Finally, the trained model is used to predict the performance of the new ultra-high voltage cable. The relevant parameters of the cable are input, and the model will output the predicted performance indicators. The prediction results are evaluated and analyzed to determine the accuracy and reliability of the prediction. The implementation steps of using genetic algorithm to optimize support vector regression SVR in step S3 are as follows: S31. The first is the coding design. The chromosome coding of the individual is divided into two parts. One part is used to represent the parameters of SVR, the penalty coefficient C and the kernel function parameters, using real number coding; the other part is used for feature selection, using binary coding, each bit corresponds to a feature, 0 means not to select the feature, 1 means to select the feature; S32, then randomly generate a certain number of individuals to form an initial population, each individual contains both the initial value of the SVR parameter and the initial state of feature selection; S33, then define the fitness function, which takes into account both the prediction error and the number of selected features. Suppose the data set has n samples and the actual value is y i , the predicted value is The number of selected features is m, and the total number of features is N, then the fitness function is defined as Where MSE is the mean square error, ω1, ω2 are weight coefficients used to balance the importance of prediction error and number of features; S34, the next step is genetic operation. In the selection operation, the roulette wheel method is used to select excellent individuals to enter the next generation according to the fitness value of the individuals; in the crossover operation, arithmetic crossover is used for the SVR parameter part, and single-point crossover is used for the feature selection part; in the mutation operation, real number mutation is performed on the SVR parameter part, and binary mutation is performed on the feature selection part; S35. Finally, steps S33 and S34 are repeated until a termination condition is met. The termination condition is that a preset number of iterations is reached, and finally the optimized SVR parameters and feature selection results are obtained.

2. The performance prediction method of an ultra-high voltage cable according to claim 1, characterized in that: The specific steps of using the density-based isolated point detection method to clean the collected data in step S2 are: S21. First, for each data point x i , define its k distance d k (x i ) is the distance between the point and its kth nearest neighbor, specifically d k (x i )=distance(x i ,x j ), where x j is x i The kth nearest neighbor of S22, then calculate the reachable distance, for a data point x i and the point x in its neighborhood j , the reachable distance is RD k (x i ,x j )=max(d k (x j ),distance(x i ,x j )), where d k (x j ) is the point x j k distance, distance(x i ,x j ) is the point x i and x j The actual distance; S23, then for the data point x i , whose local reachable density LRD k (x i ) is defined as point x i The reciprocal of the average of the reachable distances of the k nearest neighbors Where |N k (x i )| is point x i The number of k nearest neighbors, RD k (x i ,x j ) is the point x i and its neighboring point x j The reachable distance; S24, then calculate the local outlier factor to measure the data point x i The difference between the density of and the density of other points in its neighborhood Among them LRD k (x j ) is the point x j The local reachable density of S25, according to the value of the local outlier factor, set the threshold, when LRD k (x j )≥3, it is marked as an outlier.

3. The performance prediction method of an ultra-high voltage cable according to claim 1, characterized in that: The operation steps of using the trained model to predict and evaluate the performance of the new ultra-high voltage cable in step S4 are as follows: S41. First, for the new ultra-high voltage cable data, its performance results are predicted through the model and its real performance index data is obtained through actual testing or measurement; S42, then use the absolute error to calculate the error, and use the mean absolute error and the root mean square error as evaluation indicators to evaluate the accuracy of the prediction; S43. Finally, the prediction results are evaluated and analyzed. According to the values ​​of the error index and the evaluation index, the prediction performance of the model is analyzed. If the error index exceeds the threshold of 8%, the model is further optimized by adjusting the model parameters, increasing the amount of data, and improving the feature selection. Different types of ultra-high voltage cables or different parameter ranges can also be analyzed separately to determine the performance of the model under different conditions.

Images