直流电机电刷磨损预测的粒子滤波方法

A Particle Filter Method for Brush Wear Prediction of Direct Current Motor

针对直流电机电刷磨损的跟踪与预测问题,提出了一种改进型粒子滤波器的电机电刷磨损状态演变预测方法。首先建立直流电机的动态模型,通过电刷磨损仿真分析获得一定时间间隔的电流输出数据,此输出数据作为测量数据用作为预测模型的输入。接着,根据测量数据的变化规律建立电机电刷磨损预测的模型形式,然后通过采用所提出的多项式重采样粒子滤波方法来预测电机电刷磨损状态的演变。案例数据分析结果表明,所提出的基于粒子滤波的电刷磨损预测方法是有效的且计算过程简便。

Aiming at the issue of brush wear tracking and prediction of DC motors, a revised particle filter method was proposed for predicting the evolvement of brush wear condition of DC motor. Firstly, a dynamic model of DC motor was built by considering the influence of motor’s commutation. The current output data with some time length were obtained by brush wear simulation. These current data were considered to be measuring data and used to be input of the designed predicting model. Secondly, the model version for brush wear prediction was established according to the changing trend of the measuring data. Eventually, brush wear prediction was performed by the particle filter with multinomial resample based on the measuring data. The analysis results of case data show that the proposed brush wear prediction method based on particle filter is effective and its calculation process is straightforward.