Neural Networks for Synthesizing Linear Feedback Control Systems Via Pole Assignment

Artificial neural networks involve a network of simple processing elements (artificial neurons) which can exhibit complex global behavior. This behvavior is determined by the connections between the processing elements and element parameters. Artificial neural networks can be easily trained to perform a particular function by adjusting the values of the connections (weights) between elements. In this manner, neural networks provide sophisticated and efficient information processing which makes them applicable for solving broad range of artificial intelligence problems. Becouse of ability to solve problems that are difficult for solving for conventional computers or human beings (speech recognitionn, image analysis, classification, adaptive control, autonomus robots control...) neural networks have been applied in almost all branches of engenering systems. One of the applications is in control systems where artificial recurrent neural networks provide an effective on-line dynamic approach for synthesizing linear control systems via pole assignment. To ensure control system stability recurent neural network performs self tuning of control parameters as respond to system parameters changing. This kind of control system is called adaptive control system. If the state variables measuremant is not possible or it is very slow or expensive neural network can be used for system identification, in other words, for determination of unknown state variables.