Multi-Agent Genetic Algorithm And Its Application In The Design Of Full Dimension State Observer

The theory and practice of control system, brings a great influence for the production of human activities and social activities. Since the early 1960s, with Kalman (R. E. Kalman) the conception and method of state and state space system into the system control theory, greatly promoted the theory of linear systems in time domain. Compared with the classical control theory, carries on the analysis system of comprehensive, modern control theory is used to describe the system state variables within the physical characteristics of the system. State feedback correction can provide more information.However, the state as a system of internal variables, in most cases, are not all directly measured. many measurement means, due to the limitation of economy and the applicability of the state feedback physical realization become impossible or very difficult thing. But in control engineering, the state feedback can not be replaced in the correction, disturbance rejection and decoupling control, and the two aspects forms a great contradiction. Therefore, to promote the state reconstruction is the formation and development of the theory of state observer.Through the research on the state observer, the use of the controlled system and the status of observer output error objective function, through a multi-agent genetic algorithm to optimize the objective function. The agent has the perception of the environment and by their own autonomous intelligent reaction to environmental capacity, parallel genetic algorithm with global searching ability we can make full use of the information within the population, to avoid the genetic algorithm into local optimal solution. In this paper, the final results of experiments show that the algorithm can more accurately determine the status of observer output feedback design matrix.