Parameter Identification Of Quantized Control Systems

In industrial control, performance of real-time and accuracy of signal transmission have effect on system stability and performance. Control systems compose by some complicated control processes in modern industrial. Considering the limitation of channel capacity of control process, the data quantized processing is necessary, which main advantage of is to reduce the total number of information quantity transmitted through system channels, so as to save the limited communication resource and to facilitate the expansion and monitoring of the control system. With computer, network and other digital equipments widely used in modern control systems, a class of communication constrains system between controller and controlled object forms. In such systems, a number of channels contained in control cycle only limited real-time data transmission, so the impact of communication constraints can't be ignored during the system analysis and synthesis, especially in parameter estimated and system identification. System output signal inaccurate under quantization, in addition to coupled with measurements noise, it makes more difficult to parameter identification, so research on parameter identification of quantized control system have important theoretical and practical significance.This paper studies identification of quantized control systems subjected to communication constraints firstly. An auxiliary model based identification method for quantized systems is introduced by employing repeated stochastic empirical measurements. The system features and a two-step identification strategy are presented. It is shown that the quantized system based on repeated stochastic empirical measurements involves time-varying estimation error, and the persistent exciting condition for parameter identification is derived. The auxiliary model based quantized multi-innovation recursive algorithm for quantized system is also given then. Convergence analysis of the auxiliary model based algorithm provides an upper bound computation way for parameter identification error estimation. It is demonstrated that under some certain conditions, recursive algorithms is consistently convergence. Simulation results show the effectiveness of the methods.Secondly, to the control systems with dual-rate sampling and signal quantization, an auxiliary model based system identification method for dual-rate sampling and quantized systems is presented by employing repeated stochastic empirical measurements. The model features of dual-rate sampling system and a two-step identification strategy are presented under relaxed estimated error conditions. It is shown that the dual-rate quantized system based on repeated stochastic empirical measurements involves time-varying estimation error, and the persistent exciting condition for parameter identification is derived. The auxiliary model based quantized recursive identification algorithm for dual-rate sample quantized systems is also given then. Convergence analysis of the auxiliary model based quantized identification recursive algorithm provides an upper bound value for parameter identification error estimation. Finally, simulation results show the validity of the conclusions.Finally, to the single-rate sampling and dual-rate sampling Hammerstein nonlinear systems with signal quantization, auxiliary model based system identification method for Hammerstein quantized systems are presented by employing repeated stochastic empirical output measurements respectively. The model features of output signal quantized Hammerstein nonlinear system and a two-step identification strategy are presented under relaxed estimated error condition respectively. The persistent exciting conditions for parameter identification are derived. The auxiliary model based quantized parameter recursive identification algorithm for Hammerstein nonlinear systems are also given then. Convergence analysis of algorithms provides an upper bound value for parameter identification error estimation respectively. Simulation results show the effectiveness of the conclusions.