Wheel Load Estimation and Anti-Roll Bar Control Using Suspension Analysis with Neural Network

Automotive by-wire chassis technology is one of the crucial technologies for intelligent vehicles, which currently includes various types of actuators including active suspension, electronic anti-roll bar, etc. The control of the above actuators relies on accurate dynamic data, of which wheel load is the one. However, measuring wheel load with high accuracy has been a challenge. In this paper, the wheel load is calculated by suspension analysis and the algorithm is fitted with neural networks to realize fast computation on board. The measured wheel load is also used to control the anti-roll bar successfully and improve the cornering stability. The innovations of this paper include determining the wheel load without the exact model of the whole vehicle, which enhances the application scope; building Genetic Algorithm (GA)-based Long Short-term Memory (LSTM) neural network, which enhances the computational speed; maximizing the use of the existing sensor data, which reduces the hardware cost. The results of the experiment on a Formula Student car meet expectations.