基于粒子滤波与负向选择算法的GIS设备故障检测

为了及早发现故障合理安排设备检修计划，提出一种基于粒子滤波与负向选择算法的GIS故障检测方法。首先，选取GIS设备金属外壳振动信号分形维数作为特征变量，有效削弱了设备负载变化对外壳振动的影响。同时，基于粒子滤波及支持向量回归算法处理设备正常状态下的振动信号分形维数特征样本，建立GIS设备振动特征估计器。将实时测量的振动特征输入特征估计器，计算估计器输入值与输出值之间的残差并作为检测指标。最后，利用负向选择算法处理正常状态下检测指标数据，间接获取GIS故障状态下检测指标区间，进而实现设备故障的检测。通过对现场实际测量数据的仿真分析，验证了所提方法的可行性。

Fault Detection of Gas Insulated Switchgear Based on Particle Filter and Negative Selection Algorithm

In order to find the fault reasonably and arrange the equipment maintenance plan early, a GIS fault detection method based on particle filter and negative selection algorithm is proposed. Firstly, the fractal dimension of the vibration signal of the metal case of the GIS equipment is selected as the characteristic variable, which effectively reduces the impact of the change of the equipment load on the vibration of the case. At the same time, based on particle filtering and support vector regression algorithm, the fractal dimension feature samples of vibration signals under normal state of the equipment are processed to establish a GIS equipment vibration feature estimator. The vibration characteristics measured in real time are input to a feature estimator, and the residual between the input value and the output value of the estimator is calculated as a detection index. Finally, the negative selection algorithm is used to process the detection index data in the normal state, and the detection index interval in the GIS fault state is obtained indirectly, so as to realize the detection of equipment faults. The feasibility of the method proposed is verified by the simulation analysis of the actual measurement data in the field.