Control System Reliability Analysis And Optimization Design

In this paper, the basic theories about the reliability analysis, genetic algorithms (GA) and the realization techniques of GA are presented and reviewed briefly. The formulation of the problems of reliability analysis and reliability-based design optimization of control systems is investigated. Genetic algorithm is applied to solve this problem and the GA based method for system optimization and synthesis of the reliabilities and the redundancies of the components and/or sub-systems is presented. The optimal allocation designs of the reliability and redundancy of the serial-parallel system and complex system are given. The main contents includes:1. The Bathtub curve-like failure rate model of the components and systems has been summarized and analyzed. The conclusions are obtained that (1) the mixed Weibull distribution can simulate the failure rate curve of components and systems; (2)The system reliability can be improved by simplifying the system structure and improving the reliability of components and subsystems; (3) Increasing the redundancy of the components is the main way to improve the system reliability; (4) The reliability-based system design method, in which the reliabilities and the redundancies of components and/or subsystems are designed and synthesized by optimization techniques, is the important design method of improving the system reliability.2. The basic theory and encoding technique of the genetic algorithm are reviewed. Considering the special character of the reliability design optimization of control system, the integer coding technique is suitable for discrete variable problems and the mixed coding method such as binary and symbol coding, and binary and floating coding strategies can be used to optimize the allocation design of the component reliability (continuous value) and redundancy (integral, discrete value) of systems.3.Applying the reliability concept and reliability based analysis and optimization method to the system design, the mathematical formulation of the reliability based system optimization design problem is made. In this formulation, the reliability, redundancy and fault-tolerant of both the hardware (components) and the software are designed simultaneously to make the reliability of the system maximized or the cost minimized. By applying the genetic algorithm to the system reliability design, a genetic algorithm based method for the component's reliability, redundancy andfault-tolerant distribution optimization of systems is proposed. By comparing and analyzing the results obtained from the optimization of the component type choosing and the redundancy, under the limits to the cost and weight of the simple serial-parallel systems with the results obtained by traditional method, the validity of the genetic algorithm applied in such questions was proved. A numerical result of an optimization design based on reliability and redundancy subject to cost in complex system is provided in the paper.