Closed-Loop System Identification in Continuous-Time Using Parallel PI Controller and Reference Prefiltering:White Noise Case

Some industrial processes require closed-loop system identification due to constraints on direct system access, leading to challenges with cross-correlated signals. Instrumental variables can estimate noise-free control and output signals. An optimal filter minimizes noise while preserving essential signal information. In Computer Numerical Control (CNC) machining, the spindle speed control loop involves a parallel proportional-integral controller with specific prefiltering for the integral component, and consequently the classic CLSRIVC method does not work properly, even if the output is affected by white noise. This paper investigates various closed-loop identification methods under white noise conditions and evaluates their statistical properties using Monte Carlo simulations.