Dynamic Neural Network-Based Feedback Linearization Control of Antilock Braking Systems Incorporated with Active Suspensions

The paper presents the design of a nonlinear dynamic neural network-based feedback linearisation (DNNFBL) control scheme for an antilock braking system (ABS) incorporated with a half-car active vehicle suspension system (AVSS). Highly nonlinear mathematical model for the integrated ABS/AVSS was derived from first principles. The dynamic neural networks were trained offline using the quasi-Newton algorithm. Simulation results reveal that the DNNFBL controller exhibits faster response times, greater power being applied to the actuators, and reduced braking time and distance in comparison to the PID and PDF-controlled systems.