基于辨识算法的风力机桨距执行器故障诊断

Fault Diagnosis for Pitch Actuators of Wind Turbines Based on Identification Algorithm

针对风力机桨距的系统故障问题,提出一种基于变遗忘因子递推最小二乘算法(Variable Forgetting Factor Recursive Least-Squares,简称VFF-RLS)的故障诊断方法。根据桨距执行器故障会引起系统参数变化的特点,采用该算法对变化的参数进行估计,将执行器故障诊断问题转化为参数估计问题。桨距执行器模型经过离散转化为系统辨识模型,进而实现对时变的桨距执行器自然频率和阻尼系数进行辨识估计,且通过自动调整遗忘因子大小保证了辨识算法的收敛速度和辨识精度。仿真结果表明,所提出的方法能够有效诊断桨距执行器故障。

In light of pitch system faults of wind turbines, a fault diagnosis method is proposed based on the variable forgetting factor recursive least-squares algorithm. According to the characteristic that pitch actuator faults can lead to the change of system parameters, the algorithm put forward is adopted to estimate the variable parameters. The actuator fault diagnosis problem is transformed into the parameter estimation issue. The pitch actuator model is transformed into the system identification model through discretization. Then the time-varying natural frequency and damping ratio of pitch actuators are estimated based on the model. The convergent speed and identification accuracy of the identification algorithm can be guaranteed by adjusting the forgetting factor automatically. The simulation results demonstrate that the proposed method is able to diagnose the pitch actuator faults effectively.