Disturbance Rejection using Neuro-Fuzzy Controllers

Non-linear uncertain systems are difficult to model and control especially when the system has significant amount of disturbances. Achieving tight control performance in the face of poorly measured disturbance is a difficult objective. To reject the disturbances different methodologies have been proposed. The advantages and disadvantages of Internal model Control (IMC) are discussed. Hybrid control schemes has been proposed which incorporate the feedforward of the measured disturbance along with IMC. The benefits of Fuzzy Relational Model in the representation of the measurement uncertainty constitute a major portion of the book. A novel approach to defuzzification is explored which can significantly reduce the control activity. A novel two-stage approach is proposed which is able to provide a better representation of the measurement uncertainty in the fuzzy control signal and consequently less control activity. Artificial Neural Network model of the laboratory air handling unit has been developed. Results from the simulation as well as from the real system are included.