System identification method with denoising and disturbance-rejecting capability

To overcome the disadvantageous influences of additive noises and temporary disturbances to system identification, system response involving additive noise is decomposed, denoised and reconstructed multiply by employing different wavelet functions and different threshold-based methods, then the identification results are obtained by employing the classes of LS (least square) identification algorithms to the denoised identification data. However, LS Algorithms are quite sensitive to temporary disturbances and lack of effective disturbance-rejecting capability. Thus, an identification algorithm based on the criterion function of least absolute error (LA) is presented, and the corresponding solving approach is also given. Unlike the classes of traditional IS identification methods, the presented LA identification method can reject heavy temporary disturbances and lead to good identification results.