基于支持向量机的轧钢电机状态综合评价方法

A Method for Rolling Large Motor State Evaluation Based on Support Vector Machine

提出了一种面向设备管理的基于支持向量机(SVM)的轧钢电机状态综合评价方法。选取与电机状态密切相关的参数为评价指标,构建模拟电机状态综合评价体系的SVM评价模型,输出电机状态综合评价等级,制定绝对报警和相对报警两种警情设置,并建立警情溯源机制,从而为及时准确掌握电机状态和预报电机状态迁移趋势提供依据,有效防范渐变性故障的发生。以济钢热连轧厂粗轧上位电机为评价对象,基于历史监测数据构建电机状态SVM评价模型,并检验其有效性。实验结果表明,粗轧电机状态评价方法整体准确率为96.67%,对异常设备状态捕捉率为88.89%,异常状态误报率为3.33%。

An equipment management oriented method for Rolling Large Motor State Evaluation Based on Support Vector Machine（SVM）is proposed.The method selects the state-related parameters as the evaluation index,and then builds the SVM evaluation model of the motor state which gives the comprehensive evaluation levels.The state levels can help understand the exact motor state even the state trend that can guard against equipment malfunction.To verify the effectiveness of the evaluation method,the experiment for the top rough roll motor is conducted which trains the SVM evaluation model based on the historical monitoring data.According to the results,the total accuracy of the evaluation method is 90%,the capture of the abnormal state is 96.67%,and the false alarm is 6.67%.