多重分形去趋势波动分析及改进决策树在电能质量分析中的应用

通过多重分形去趋势波动分析方法分析了6种常见的电能质量信号,证明了电能质量信号具有多重分形特征。据此提出基于多重分形去趋势波动分析的电能质量特征提取方法,选取多重分形谱参数(h_qmax、α_min、α₀)和信号能量E作为特征向量矩阵,结合改进决策树分类,进行电能质量分析和识别。该方法与DTCWT、HHT和EEMD方法进行对比实验,结果表明,该方法表现出更好的识别结果,为电能质量信号的特征提取提供了一种新的思路。

Application in power Quality Analysis Based on Multifractal Detrended Fluctuation Analysis and Improved Decision Tree

Six common power quality signals are analyzed by multi-fractal and trend fluctuation analysis, which proves that the power quality signal has multiple fractal features. Based on this, a power quality feature extraction method based on multi-fractal detrended wave analysis is proposed. Multi-fractal spectrum parameters (h_qmax、α_min、α₀) and signal energy E are selected as feature vector matrix, combined with improved decision tree classification for power quality. Analysis and identification. The method is compared with DTCWT, HHT and EEMD. The results show that the proposed method shows better recognition results and provides a new idea for feature extraction of power quality signals.