摘要
基于人工神经网络在非线性系统辨识中的优越性,依据整车的各项参数、动力性及经济性的试验测试结果,分析了系统辨识精度的评价指标,确定了两隐层BP神经网络的神经元数、初始权值及初始阈值,建立了汽车性能数据库的神经网络模型.该模型可实现正向预测与逆向推理功能,可缩短汽车的研发周期,并利用实例进行了说明与验证.
Based on the superiority of artificial neural network （ANN） in identification of non-linear system, according to parameters of automobile, dynamic performance and fuel economy gained by experiment, evaluation criterion of system identification accuracy is analyzed. Furthermore, initial weights, initial biases and nodes of BP ANN has two layers are gained. Then, performance database of automobile is established. This model can realize forward prediction and backward reasoning, and it can decrease the development cycle of automobile. At last, validation is carried out by means of example.
出处
《重庆大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2006年第1期8-11,共4页
Journal of Chongqing University
基金
上海汽车工业总公司资助项目(0308)
上海市教委曙光计划资助项目(02SGU8)
关键词
人工神经网络
汽车性能数据库
建模
artificial neural network
automobile performance database
modeling
