Soft computing based networking technique and its application to control system design and identification

Modelling and designing a controller of a nonlinear plant is particularly difficult when the plant to be controlled is significantly nonlinear. Forming a nonlinear model from a fundamental equation may not be possible in some cases and designing a controller for it is as difficult. To address this problem and ease the difficulties in modelling and designing of industrial controllers, this paper proposes a trajectory controller network (TCN) and linear approximation model (LAM) technique. TCN and LAM proposed are formed by multiple linear blocks and both can be obtained directly from plant step response data using evolutionary computation which offers a powerful and practical alternative to a wide variety of traditional engineering, mathematical and statistical problem-solving techniques