A non-parametric approach for design of intelligent control system

In this paper, a methodology using neural network and genetic algorithm for failure detection and smooth accommodation is presented. The main idea is to constantly monitor the system input and output to identify any off-nominal (failure) conditions using two levels of intelligence decision-making algorithm. The genetic algorithm is used to do online search for near-optimal coefficient values for combining two controllers (before and after failure) which provide smooth accommodation of the failure. Preliminary research findings showed that the proposed intelligent control system can detect and accommodate single failure modes and generate an appropriate control action for this mode. Simulation results, conducted on the control of an advanced model aircraft showed that this intelligent control system is able to maintain the stability of the system even in case of harsh failures in the dynamics of the airplane.