基于时序贴近度与改进SVM的水机轴心轨迹诊断

Axis orbit automatic identification on time-series similarity mining for hydropower units

为了提高水轮机组诊断的精确性,提出应用时间序列模糊贴近度特征提取轴心轨迹特征参数,通过改进SVM模型并引入故障分类准确性判定因子对参数化的水电机组轴心轨迹开展了智能诊断.应用改进SVM对时间序列特征引入正确率、错误分类率计算方法,从而对诊断后轴心轨迹分类准确性进行判定,由此促进运行状态设备智能诊断,提高故障诊断系统的自动诊断水平及准确率;引入多类分类支持向量机算法、分类准确度判断解决异常状态下机组轴心轨迹特征参数无法识别、识别率低的问题.通过对改进扩展时序距离时间序列贴近度度量算法的应用解决了水电机组实时轴心轨迹特征参数准确性差和实时性差的问题.该方法提高了检测精度,同时增强了人机交互性,具有重要的理论意义和实用价值.

In order to improve the diagnostic accuracy of turbine units,a new method for intelligent diagnosis of shaft orbit of hydro generator set is presented.This method according to the characteristics of the hydroelectric generating unit,the application of the fuzzy distance measure between features with recent time-series,the application of improved intelligent diagnosis on time sequence characteristics of SVM,improve the diagnostic level of automatic fault diagnosis system and accuracy.The results show that axis orbit automatic identification on time-series similarity mining to fault diagnosis of hydropower unit was feasible,and could improve and enhance intellectualize and humanization of diagnosis,as well as the human-computer interaction.