Strategies for genetic adaptive control

In this paper, we investigate ways to use genetic algorithms in the online control of a nonlinear system and compare our results with conventional control techniques. We develop a direct genetic adaptive controller, an indirect genetic adaptive controller, and combine the two into a general genetic adaptive controller. We also examine several conventional controllers including a proportional-derivative (PD) controller, a model reference adaptive controller, and two indirect adaptive controllers. To demonstrate all these control techniques, we investigate the problem of cargo ship steering. In this application, we describe the desired performance with a reference model and use our control techniques to track the output of the reference model. Overall, our goal is not to design the best possible controller for ship steering; we simply use this example to illustrate the ideas.