为了提高复合电能质量扰动(power quality disturbance,PQD)识别结果的正确率,提出了一种基于改进灰狼优化(im-proved grey wolf optimization,IGWO)算法核极限学习机(kernel extreme leavning madine,KELM)的复合PQD识别方法.利用S变换获得复合PQD信号的特征量,以此作为复合PQD识别模型的输入量.采用精英反向学习、自适应收敛系数和柯西变异这3种策略对灰狼优化算法进行改进,得到全局搜索性能更好的IGWO算法.采用IGWO算法对KELM的核系数和惩罚参数进行优化,建立了基于IGWO-KELM的复合PQD识别模型.仿真分析结果表明,该模型识别的准确率高达98.10%,识别效果明显优于其他方法.
Composite Power Quality Disturbance Identification Based on IGWO-KELM
In order to improve the accuracy of composite power quality disturbance(PQD)identification,a method based on the improved grey wolf optimization(IGWO)-kernel extreme learning machine(KELM)for composite PQD identification is pro-posed.The characteristic parameters of composite PQD signals are obtained using the S-transform,which serves as the input of the composite PQD identification model.The grey wolf optimization is improved by using elite reverse learning,adaptive conver-gence coefficient,and Cauchy mutation to obtain the IGWO with better global search performance.The IGWO algorithm is then applied to optimize the kernel coefficient and penalty parameter of KELM,and the composite PQD identification model based on IGWO-KELM is established.Simulation results show that the proposed IGWO-KELM model is superior to other methods signifi-cantly in composite PQD identification with an accuracy rate of 98.10%.
composite power quality disturbanceidentificationimproved grey wolf optimizationkernel extreme learning ma-chineaccuracy rate
NETL
NSTL
万方数据
