协同滤波器和支持向量机在HVDC系统故障诊断中的应用

Application of Consensus Filter and SVM in Fault Diagnosis for HVDC Systems

提出一种基于协同(consensus)滤波器和支持向量机的HVDC系统故障诊断方法。由于随机噪声的干扰,测得的输电线路中直流电压信号无法直接用于故障的检测,使用协同滤波器对多个传感器测得的直流电压信号进行滤波,然后将滤波后的结果用来构建故障检测滤波器,检测故障的发生。为更好的对检测出的故障进行分类,有效提取故障特征,对直流电压波形进行S变换。变换后的特征量作为支持向量机(support vector machine,SVM)的输入,建立系统故障诊断模型,并比较不同参数下的SVM模型性能。仿真结果表明,故障检测有效,且对不同的故障能正确的进行分类。

A fault diagnosis scheme for HVDC （high voltage direct current transmission）system is proposed based on the consensus filter and support vector machine theory. The measured volts d.c. signals in the DC line can not be used to detect system falut on account of the effect of the random noise. Consensus filter is used to filter the volts d.c. which are measured by multi-sensors. A fault observer is constructed by using the filtered volts d.c. signals. In order to classify the detected faults and extract the faults features effectively, volts d.c. signals are selected to be transformed by using Stransformation method. The extracted volts d.c. features are used to train SVM, and then the SVM performance with different parameters is compared. Simulation results show the efficiency of the proposed method.