Research On Quantized Estimation And Quantized Control In Networked Control System

Networked control system is a combine of network and control technology, it has been generally used in industry, economic and other aspects, which brings a new interest of study. As there are not only stochastic noises but also time delay and quantization that can influence the performance of networked control system, so in order to understand the networked control system better, we have to consider all the factors synthesized. This paper studies quantized estimation of the networked control system, the mini-mum bit rate of stochastic system with multiplicative noises and system with time delay in the input channel respectively, the quantized control for system with quan-tized performance and further study of the date bit of the system for it to be stabilized. The main works and contributions are summarized as follows:1. For general stochastic system and stochastic control system with multiplica-tive noise and addictive noise, using the sector bound approach to characterize the quantization error using a norm bounded multiplicative noise and the innovation, us-ing the mean square techniques, we design performance guaranteed quantized filter by solving Riccati equations.2. The problem of minimum capacity for stochastic system with multiplicative noise to be mean-square stabilized is considered. The coarsest quantizer is derived, and the solvability of the corresponding quantization density and a linear quadratic regulator problem are established, then the problem of finding the coarsest quantiza-tion is transformed into solving an algebraic Riccati equation, which can be solved by linear matrix inequality to get a suboptimal quantization density. By searching over the parameters, the optimal quantization density can be approximated by the suboptimal quantization density. The results have also been generalized to the mini-mum mean-square stability with exponential index.3. For the problem of the coarsest quantization density of stochastic system sub- ject to multiplicative and additive noises to be mean-square stabilized via quantized control, a quantization density related Lyapunov function is proposed to studied the stability of the system. By using linear matrix inequality, sufficient and necessary conditions are given for the system to be quantized stabilized, and the quantization density is smaller than the one deduced by common Lyapnuov function.4. For the minimum mean-square stabilization of single input system which subject to time delay, by using augmentation to build a new system without time delay. Deducing the accessible control set of the two systems, an isomorphism be-tween the two control sets is built, the problem of finding the coarsest quantization density for the system subject to input-delay is transformed into deducing the coars-est quantization density for the new built system without time delay. By using the approach introducing chapter2, the coarsest quantization density for the system to be stabilized is characterized by the unstable roots of the system matrix.5. For quantized stabilization problem with performance index of stochastic system with multiplicative noises, conditions for the existence of the solutions are given. The relations between the solutions of quadratic quantized control and guar-anteed cost control problem are built, the solutions to the problems are given by linear matrix inequality, which supplies a toll for the quantized control problem with performance index.In conclusion, this dissertation focuses on the quantized control and quantized estimation problem in networked control system. The obtained results have not only important theoretic values, but also extensive practical values.