论文部分内容阅读
针对电能质量识别领域中,采用随机参数的支持向量机(SVM)分类器识别随机暂态扰动信号准确率低、优化耗时长等问题,提出一种基于遗传算法(GA)优化SVM识别电能质量暂态扰动(PQD)的新方法(GASVM)。首先,仿真生成具有随机噪声水平和扰动参数的9种PQD信号;接着,通过S变换,提取出6种信号特征构成输入特征向量,用于训练SVM分类器;再采用GA对SVM进行参数寻优,进而获得优化的GA-SVM分类器;最后,采用GA-SVM识别PQD信号。仿真对比试验表明,新方法能准确识别不同噪声环境下的9种PQD信号,分类准确率及优化所需时间均优于PSO优化SVM方法(PSO-SVM)。
In the field of power quality identification (SVM), SVM classifier is used to identify problems such as low accuracy and long time-consuming optimization of random transient disturbances, and a genetic algorithm (GA) based SVM is proposed to identify power quality New Approach to State Perturbation (PQD) (GASVM). First of all, nine kinds of PQD signals with random noise level and perturbation parameters are generated by simulation. Then, six kinds of signal features are extracted by S transform to form input eigenvectors, which are used to train SVM classifier. Then GA is used to optimize parameters of SVM , And then get the optimized GA-SVM classifier; Finally, using GA-SVM to identify PQD signal. The simulation tests show that the new method can accurately identify nine kinds of PQD signals in different noisy environments, and the accuracy of classification and the time required for optimization are better than PSO-SVM (PSO-SVM).