Search

Owing to the limitation of production cost, the idle shake problem is very common among minivans. To improve the performance of the steering system while keeping the cost not increased, an enhanced hybrid and adaptive meta-model based global optimisation (E-HAM) is applied in this work. Based on the HAM algorithm, the new important region update method (IRU) and different sampling size strategies are integrated in the E-HAM. The resulting precision, convergence rate and search efficiency are improved remarkably. The minivan steering system is then optimised by another six optimisation algorithms to perform the comparisons. Finite element method (FEM) analysis and experiment are also conducted to verify the effectiveness. The results show that the structures are optimised successfully by the E-HAM optimisation method and the idle shake problem is well solved. The research in this work also serves a good example for other structure optimisation problems.

To solve the problem of trajectory tracking for 4WS autonomous vehicles under high speed emergency obstacle avoidance, an improved trajectory tracking controller combining four-wheel steering (4WS) feed-forward control and linear quadratic regulator (LQR) feed-back control is proposed. First, a polynomial obstacle avoidance trajectory of small radius and sharp curve is established and its several variables are analysed. Second, in order to make actual yaw rate follow steady-state value based on reference model, the steering gain of the improved 4WS feed-forward control can change with steering angular frequency and vehicle speed. Third, combined with the LQR feed-back control, the deviations between actual lateral displacement and desired lateral displacement are controlled under 0.05 m. Finally, CarSim® simulation results show that 4WS autonomous vehicles based on the controller can not only avoid obstacles in time, but also improve active safety and handling stability under high speed emergency obstacle avoidance.

A soft computing arithmetic based on Adaptive Neuro-Fuzzy Inference System (ANFIS) is proposed to estimate vehicle yaw rate in a driver-vehicle closed loop system, and vehicle yaw rate is considered as a nonlinear mapping of time series of rack displacement and lateral acceleration. Simulation study on soft computing of vehicle yaw rate in the driver-vehicle closed loop system is conducted, and the performance of the soft computing arithmetic based on ANFIS is evaluated with the help of actual vehicle test data. Results indicate that the generalisation of the soft computing arithmetic based on ANFIS is better than that based on Radial Basis Function (RBF) neural network.