基于自适应FIR预测滤波器的谐波检测

针对现阶段有源电力滤波器畸变电流检测方法存在工频周期时延、计算量大等不足的问题,提出了基于自适应有限脉冲响应(FIR)预测滤波器的谐波实时检测系统。论述了自适应滤波器谐波检测原理并利用变步长的最小均方算法(LMS)对所需检测信号进行预测,而预测算法的步长因子是根据误差信号的时间均值估计来调节的,即当滤波器的预测系数远离最优解时,步长比较大,以加强动态响应速度和对时变系统的跟踪能力;当滤波器的预测系数接近最优解时,步长比较小,以获得较小的稳态误差。对该预测法采用MATLAB进行了仿真和实验,结果表明当电流突变时,该方法仍然能够在一个周期内正确预测出未来时刻的谐波电流值。

Harmonic Detection Based on Adaptive FIR Predictive Filter

The dynamic compensation of active power filter(APF)demands that variable current in electric power network should be detected exactly and real-timely.To avoid some shortages of current harmonic detection methods of APF,a real-time detection system based on adaptive finite impulse response(FIR)predictive filter is presented.The harmonic detection principle is discussed,and the variable step-size least mean square(LMS)is used to predict the detected signals,which uses the estimation of the time-average of error signal to control the updating of the step-size.When the predictive coefficient is far away from the optimal value,it increases the step-size to accelerate the tracking of time-variable system,while when the predictive coefficient is near the optimal value,it decreases the step-size to minish the steady state error.The results of the simulation and experiments through MATLAB simulation software indicate that in spite of the current break,the method can also correctly predicate the future harmonic current value in a certain period.