【摘要】 分类特征合理选取和分类方法是短时电能质量复合扰动分类的核心内容。以科学分类的可比性原则为出发点,引入分类特征概念,提出短时电能质量复合扰动分类特征3层选取策略,以主要频率点、幅值特征等为分类特征,研究分类特征提取算法。在突出不同类扰动分类特征差异性的同时,考虑同类扰动特征的相关性,提出基于马氏距离的短时电能质量复合扰动分类方法。对8大类单一短时扰动及其构成的复合扰动共16种进行仿真,并和支持向量机神经网络2种分类器进行比较,实验表明提出的方法实时性好且准确性高,有一定的工程应用前景。更多还原

【Abstract】 Reasonable feature selection and classification method are the core contents of short-time power quality complex disturbances classification. Regarding comparability principle of scientific classification as the starting point and leading in the concept of classification feature, a three-layer classification feature selection strategy for short time power quality complex disturbances is proposed, and taking main frequency point and amplitude characteristic as classification features, the classification feature extraction algorithm is researched. Considering the correlation of disturbance features of the same kind, a Mahalanobis distance based short time power quality complex disturbance classification method is proposed while the diversity among disturbance classification features of different kinds is emphasized. Eight types of single disturbances and their complex disturbances, namely sixteen in total, are simulated and the comparison of the simulation results with those by two different classifiers respectively, i.e., the classifier based on support vector machine and the classifier based on neuron network, shows that the proposed method has the ability of good real time and the classification is more accurate,hasing a good application on engineering.更多还原