Minimization of actuator repositioning in delayed processes using flexible neural networks

In process control the main goal is not only the output tracking, but also the design of a suitable control signal because it is dealt with the actuators directly. This means that the undesired oscillations should be removed to alleviate the malfunction of actuators. In this paper, a nonlinear saturating element is designed with the help of an adaptive neural network. This element would chop the undesired oscillations of the control signal. The performance of the system is guaranteed by the use of a central controller such as PID and implementation of a performance measurement index to reduce the error between the output and desired input signal. The multilayer perceptron network, with its adaptive activation function is used and trained with Error Back Propagation algorithm. The simulations applied to a Heating, Ventilating and Air Conditioning (HVAC) system, demonstrates the good performance of the new design method.