System Identification in Adaptive Inverse Control for Electrohydraulic Shaking Table Using Variable Tap-Length NLMS Algorithm

摘要

Adaptive inverse control (AIC) is an advanced control technology which can efficiently learn and control the severely nonlinear and real-time plant, such as the electrohydraulic shaking table (EST). The foundation of AIC is the system identification of the plant via an adaptive filer. The traditional adaptive filter theory is known as a fixed tap-length (length of the weight vector) FIR (Finite impulse response) filter whose weights change by adaptive algorithm automatically. The mismatch between plant and the fixed tap-length will cause inaccuracy or compute complexity. This paper introduce a variable tap-length NLMS (VTLNLMS) algorithm to find the optimum tap-length (OPTL) of the model of multi-DOF EST, then use AIC to improve the accuracy of time wave replication. The dynamics response nonlinear model of the EST is built to reflect the experimental situation. The nonlinear EST model and referred algorithm are simulated in MATLAB/Simulink. Results show the AIC with OPTL is effective in the EST.