基于相似度的道岔健康状态评估及故障检测方法研究

Research on turnout health state assessment and fault detection method based on similarity

针对道岔系统故障诊断需要大量数据且较难获取的情况,提出基于波形相似度的健康状态评估及故障检测算法。使用SURF算法进行特征点预提取,提高计算的实时性;通过Hausdorff距离计算待识别曲线与标准曲线间的相似度,确定健康值及理论故障时间;对于已经处于故障状态的样本,与故障曲线库内模板曲线进行对比,选择匹配度最大的故障模板曲线所对应的故障类型作为待识别曲线的可能故障,从而提出检修意见。实例分析表明,该方法准确率高、速度快、适应性强,具有实际应用价值。

In view of the situation that the turnout system fault diagnosis requires a large amount of data and the data is difficult to obtain,a health evaluation and fault detection algorithm based on waveform similarity was proposed.Use SURF algorithm for pre-extraction of feature points to improve the real-time performance of the calculation;by calculating the similarity between the curve to be identified and the standard curve by Hausdorff distance to determine the health value and the theoretical failure time;for samples that were already in the failure state,compare it with the failure curve.The template curves in the library were compared,and the type of failure corresponding to the failure template curve with the largest matching degree was selected as the possible failure of the curve to be identified,so as to put forward maintenance advice.The example analysis shows that the method has high accuracy,fast speed and strong adaptability and has practical application value.