Dissertation > Excellent graduate degree dissertation topics show

Gradient Based Identification for Dual-Rate Sampled-Data Systems

Author: ChenXiaoMing
Tutor: DingFeng
School: Jiangnan University
Course: Control Theory and Control Engineering
Keywords: Dual-Rate System Polynomial transform Output between samples Stochastic gradient Parameter Estimation Auxiliary model
CLC: TP11
Type: Master's thesis
Year: 2008
Downloads: 63
Quote: 1
Read: Download Dissertation

Abstract

Multirate systems widely exist in industrial processes, such as chemical process control many of the soft measurement problems can be attributed to multi-rate system modeling, parameter identification, or state estimation problem. Traditional discrete-time system assumes that the input signal and the output signal of the update cycle the same sampling period, referred to as single-rate sampled-data systems. Some chemical process due to hardware constraints, the system output sampling frequency than the system slower update frequency control input is often several continuous input control signal before sampling to get a control output. Thus, in the same control system and even appeared in two groups of data at different frequencies, the corresponding system is called the double rate (multi-rate) sampling system. Seeking effective parameters of such a system identification method, has important theoretical significance and practical value. Thesis, National Natural Science Fund Project \on, through the use of dual-rate sampled-data systems polynomial transformation technique can be used to seek suitable for identification of dual-rate data model based on stochastic gradient-depth study of dual rate sampled data system parameter identification problem, obtain the following findings. 1 Use the polynomial transform technique, will lose observational data into AR System dual-rate data can be used directly for identification ARMA model, estimated that the loss of data derived ARMA system Recursive Extended stochastic gradient algorithms. Since stochastic gradient convergence is slow, in order to improve the tracking performance of the algorithm, the introduction of a forgetting factor in the algorithm, has been forgetting factor recursive extended stochastic gradient identification algorithms. In persistent excitation conditions, using stochastic process theory and martingale convergence theorem proving to mention consistent parameter estimation error converges to zero. With a simulation example shows the effectiveness of the algorithm. (2) study the dual rate sampled data ARX system parameter identification problem. Using the polynomial transform technique dual rate sampled data ARX system into dual-rate data can be used directly for recognition of dual rate ARMAX model derived estimates of such dual-rate systems Forgetting Factor Recursive extended stochastic gradient identification algorithms. Also persistent excitation conditions, the analysis of the proposed dual-rate parameter estimation algorithm is uniform convergence, simulation example shows the proposed algorithm can give a satisfactory parameter estimation. 3 studied the dual rate system with colored noise parameter identification problem. When the system noise were MA model, AR model and the ARMA model, the use of the polynomial transform technique recognition system will be converted to a dual-rate data can be used directly for identification models, each model is derived based on stochastic gradient parameter identification algorithms. Finally simulation examples were used to demonstrate the effectiveness of the various algorithms. 4 auxiliary model based on the idea of ??using a dual-rate data directly recognize dual rate sampling output error moving average parameters of the system is proposed based on the augmented auxiliary model recursive stochastic gradient algorithm, and simulation example shows the effectiveness of the proposed algorithm. With the polynomial transform technique, this method significantly reduces the computational algorithm and parameter estimation accuracy is satisfactory. Finally, a study summary and outlook, and the dual rate system, some of the difficulties facing the Institute and needs further research directions made a brief introduction.

Related Dissertations

CLC: > Industrial Technology > Automation technology,computer technology > Automated basic theory > Automation systems theory
© 2012 www.DissertationTopic.Net Mobile