基于神经网络的半主动悬架自适应模糊控制研究

在建立了五自由度车辆半主动悬架系统模型的基础上，将神经网络与模糊控制结合起来，提出一种基于神经网络的自适应模糊控制半主动悬架系统，其控制器由模糊神经网络控制器和模糊网络组成，采用快速的变斜率梯度下降算法学习，具有自适应学习功能。仿真计算表明，与被动悬架相比，神经网络自适应模糊控制性能明显优于一般的Fuzzy控制，半主动悬架系统在减小振动，提高车辆平顺性方面优于被动悬架，且车轮动载荷和悬架动挠度也得到明显改善。台架试验同样表明了半主动悬架的优良减振性能。

Vehicle Semi-active Suspension Adaptive Fuzzy Control Based on Neural Network Control Theory

Establishing a 5 DOFs of vehicle model with semi-active suspension, integrating neural network with fuzzy system, a control strategy for vehicle semi-active suspension adaptive fuzzy system using neural network was proposed. This control system consisted of a fuzzy neural controller and model network. A weight-learning algorithm was presented using the gradient descent method with a variable-slopes, which was useful to accelerate learning speed and improve convergence. Simulation results show that the proposed controller has good control performance, the semi-active suspension is superior to the passive suspension for improving the ride comfort. The experiments under different road surfaces and vehicle speeds on the test rig coincide with those of simulation quite well.