模糊神经网络在车型自动识别中的应用研究

Fuzzy Neural Networks for Vehicle Classification

提出一种基于模糊穴映射模型的模糊神经网络分类器 (FCMN NC) ,用于机动车辆的自动分类识别 ,将模糊的概念结合于网络的各层中 ,其输入是具有语义性质的隶属度 ,输出也是模糊集的隶属度。基于上述算法 ,开发了机动车辆自动识别实验系统。同时采用了一种新颖的最小二乘法LS与BP算法相结合的LSBP学习算法 ,仿真证明 。

A fuzzy neural network classifier is presented based on fuzzy mapping model,which is used for vehicle classification.The physical meaning of the weight is clear,and both of the input and output of the classifier are membership value.The experimental system of vehicle classification is introduced,and a LSBP training algorithm combined least square with BP algorithm is also put forward.Simulation shows that the learning property of LSBP is superior to BP.