基于模糊神经网络的驾驶员弯道安全感受预测模型研究

论述了在人-车-路(环境)系统中驾驶员弯道安全感受的非线性特性。为了体现车辆、道路结构对驾驶员认知过程的作用和认知结果形成的影响,驾驶员弯道安全感受模型采用弯道平曲半径、车辆运行速度和驾驶员驾龄作为道路线形、车辆结构动力性、驾驶员经验的主要特征参数集,在增加积极和消极补偿运算的基础上,构建出模糊逻辑推理规则与神经网络优化逼近运算相互结合的5层推理预测模型,并结合实例数据,进行了对比分析验证,表明5层结构模型能够较好的满足特定弯道条件下驾驶员主观安全感受的认知推理预测要求,能够准确的、稳定的再现出驾驶员主观安全感受同驾龄、平曲半径和车辆运行速度间的相互作用关系。

Forecast of Driver''''s Curve Safety Perception Based on Fuzzy-neural Network

Nonlinear properties of driver's curve safety perception are discussed in driver-vehicle-road(environment) system.Reflecting influences of vehicle and structure of road on driver cognition during driving,the structural model of driver's curve safety perception is set up.Based on the theory of fuzzy-neural network,compensated operational layer including positive calculation and negative calculation is added and 5-layer forecasting model of driver's safety perceptive reasoning model which has advantages of inference of fuzzy logic and optimized approximation of neural networks is build up too.In this model the horizontal curve radius,running speed and driving age are taken as system input variables and driver's subjective safety perception is taken as system output variable.The results of road tests in special test sections and forecasts utilizing fuzzy-neural network's reasoning operation of driver's safety perceptive model show that the relation between system inputs and system output of 5-layer forecasting model is reflected accurately and this 5-layer fuzzy-neural network model is effective to forecast driver's curve safety perception.