To enhance dynamic tracking performance and anti-disturbance capacity of finite impulse response(FIR)filters,variable discount factors are introduced to the recursive least squares(RLS)algorithm.By employing improved FIR filters to conduct modelling of industrial robot drive systems,dynamic characteristics of the target systems are identified.Then the fault detection for a target system can be utilized by analyzing the coef-ficients of the FIR filter.Finally,an application of the fault detec-tion scheme to a kind of brushless DC motor drive system is de-scribed.Compared with reference methods,the proposed scheme achieves effective fault detection and performs better in dynamic tracking and robustness according to the final simulation results.