基于BP神经网络的共享汽车仿生设计特征分析

Bionic design features analysis of shared vehicle based on BP neural network

通过仿生设计,可使共享汽车用户的心理感受获得提升。运用BP神经网络分析车身造型特征的感性评价数据,有助于找到合理的仿生设计方向。通过前期调研,选定26款被租赁率较高的共享车型作为样本,提取造型特征,构建3层神经网络模型。以不同的隐藏层神经元数对模型进行训练并分析误差,确定适宜的隐藏层神经元数、学习速率、误差范围。研究发现:BP神经网络可用于分析共享汽车造型特征的仿生设计水平,并据此获得各造型特征的仿生得分与代表性样本;基于代表性样本的特征进行创意设计,可为共享汽车的仿生设计提供依据。

Psychological feelings of shared vehicle users could be improved through bionic design.BP neural network was used to analyze sensibility evaluation data of vehicle modeling features,which was helpful to find a reasonable bionic design direction.Based on previous surveys,26 shared vehicles with high rental rate were selected as samples to extract modeling features and construct a three-layer neural network model.The model with different numbers of hidden layer neurons was trained and error analyzed to determine appropriate number of hidden layer neurons,learning rate and error range.It was found that BP neural network could be used to analyze bionic design level of shared vehicle modeling features.And then,bionic scores and representative samples of each modeling feature could be obtained.Innovative design was performed based on representative samples features.This could provide references for bionic design of shared vehicles.