基于粒子群优化算法的电缆中间头故障智能识别仿真研究

Simulation study on intelligent identification of cable intermediate head fault based on particle swarm optimization algorithm

针对电缆中间头相似较高,容易引起误判的问题,提出一种基于粒子群优化支持向量机的参数识别模型,通过局部放电电场分布图识别电缆中间头故障类型。利用Ansys软件对电缆中间头3种典型故障类型进行仿真建模,从3个维度进行局放电场特征提取,介绍了粒子群优化支持向量机电缆局放类型识别流程,电缆中间头尖刺缺陷识别准确率达100%,气隙缺陷识别准确率达86%。通过与未经粒子群优化过的支持向量机算法和BP神经网络算法进行识别准确率对比,验证该模型具有良好的泛化能力,识别准确率更高,可为后续研究改进提供基础。

To address the problem of misjudgment caused by the high similarity of partial discharge types among cable intermediate head defect types,a parameter identification model based on particle swarm optimization support vector machine was proposed,which identified the cable intermediate head fault types by using the local discharge electric field distribution map.Using Ansys software,three typical fault types of cable intermediate head were simulated and modeled,and the local discharge electric field characteristics were extracted from three dimensions.The identification process of particle swarm optimized support vector machine cable partial discharge type was introduced,the accuracy of cable intermediate head spike defect identification reached 100%,and the gap defect identification accuracy reached 86%.By comparing the identification accuracy with the support vector machine algorithm that have not been optimized by particle swarmand BP neural network algorithm,it was verified that the model had good generalization ability,higher identification accuracy,and could provide a basis for subsequent research improvement.