基于K-means和SOM混合算法的高压断路器操作机构状态评估

Status Assessment of High Voltage Circuit Breaker Operating Mechanism Based on K-means and SOM Hybrid Algorithm

为诊断高压断路器操作机构故障,文中基于分合闸线圈电流曲线,提出了采用K-means与SOM神经网络相结合的混合算法,对断路器操作机构进行状态评估。对某批次252 k V高压断路器操作机构进行分合闸线圈电流数据采集;建立了K-means与SOM神经网络相结合的混合算法模型;对测试的断路器操作机构进行状态分析。结果表明,混合算法能够将操作机构不同状态进行聚类,可将相同故障分在同一类别。并将混合算法模型与SOM神经网络模型和K-means模型作比较,结果表明,混合算法模型在计算速度和聚类准确率上都优于其他两种模型。

In order to diagnose faults of operation mechanism of high voltage circuit breaker,based on the current curve of opening/closing coil,a hybrid algorithm combining K-means and SOM neural network is proposed to assess the status of circuit breaker operation.The data of opening/closing coil current is collected from the operating mechanisms of 252 kV high voltage circuit breakers.A hybrid algorithm model of K-means and SOM neural networks is established to analyze the status of the testing circuit breaker operating mechanism.The results show that the hybrid algorithm can be used to cluster the different status of the operating mechanism,and the same fault can be divided into the same category.The hybrid algorithm model is compared with the SOM neural network model and K-means model,and the results show that the hybrid algorithm is better than the other two models in calculating speed and clustering accuracy.