Hybrid fuzzy-neural network structure for vehicle seat vibration isolation

In this paper we present one of the concepts of vehicle seat vibration isolation, using five fuzzy controllers together with neural network, combining it into a hybrid system. We used fuzzy controllers as a part of a semi-active vehicle suspension system, which consists of semi-active air spring and passive damper. System is implemented between the cabin floor and vehicle seat. Fuzzy controllers are designed to adjust the stiffness of the air spring, in order to isolate the vertical vibrations that are being caused by rough road surfaces. Each controller is tuned for different road type. Which controller is active depends on the output of the neural network, representing one of the four roads. Input data for fuzzy controller are vehicle seat deflection and velocity of the vehicle seat deflection. Measure of the quality of suspension is the SEAT value, which is calculated as a quotient of effective seat vertical acceleration and effective cabin vertical acceleration (ISO 2631-1 standard). Small SEAT-value means good isolation of the seat, regarding the vibrations. This paper presents the results obtained from both the MATLAB simulations and the experiment on the real model.