Neuro-Fuzzy Volume Control for Quarter Car Air-Spring Suspension System

Abstract

One type of vehicle suspension system is the air suspension system. One of the important functions of the suspension system is to reduce vibration transmitted to the vehicle body from road unevenness. Soft suspension results in better ride comfort and worst handling. While stiff suspension results in better handling and worse comfort. This paper investigates the performance of air suspension in terms of passenger ride comfort and vehicle handling. The performance is tested once with Fuzzy logic control and another with Fuzzy control augmented with Neuro-Fuzzy inference system. A mathematical model is developed to describe the air-spring stiffness behavior. The air-spring suspension is merged on two degrees of freedom model. The air pressure is controlled by a Fuzzy controller as done previously in the literature. On the other hand, our approach is to control the volume change by connecting the air-spring volume to two additional unequal volumes through ON-OFF solenoid valves. The solenoid valves are controlled by Adaptive Neuro-Fuzzy Inference System (ANFIS). A single-degree-of-freedom setup was built from which the required training data for the ANFIS controller was carried out. The experimental setup is developed with variable excitation input in terms of frequency and amplitudes. The acceleration of the sprung mass has been measured and stored at different operating conditions and including different air volumes. The stored data is used for training of volume controller. The controller is tested with Sinusoidal excitation then with ISO 8608 road profile. It is found that increasing the air volume reduces the sprung mass acceleration. Also, the results showed that implementing the Neuro-Fuzzy controller with Fuzzy logic control reduces the sprung mass acceleration significantly by an average value that reaches 5 cm/s2 at frequencies above 1 Hz while maintaining road holding.