基于收敛性线性粒子群优化算法的1-SVM电力设备状态分析方法

State analysis method for power equipment by using 1-SVM based on Convergent Linear Particle Swarm Optimization algorithm

根据基于图像的机器学习方法在电力设备状态分析中的应用特点,设计了包括模型学习、设备粗定位及对齐分割在内的电力设备状态识别主要技术框架。针对利用图像对关键电力设备进行对齐分割的问题,配合图像模型中的主要特征提取方法,建立了基于收敛性线性粒子群优化算法的1-SVM电力设备状态精确对齐及分割方法。仿真实验验证了该方法在以隔离开关为对象的电力设备分析识别中能够获得较快的处理速度,同时具备较好的分类性能。

Main technical framework including model learning, equipment coarse positioning and alignment segmentation is proposed to deal with the characteristics of power equipment analysis by using image-based machine learning methods. Especially, according to the alignment segmentation problem for important power equipment images, by coupling with extraction of key features of image model, Convergent Linear Particle Swarm Optimization （CLPSO） algorithm is used in 1-SVM to tackle the precise alignment and segmentation problem for power equipment. Simulation results show that in the disconnecting switches based equipment state analysis problem, the proposed method can achieve faster processing time and better classification performance during equipment classification and identification.