摘要
本文从行驶工况特征参数来进行燃油消耗的预测。首先对典型道路上采集的百公里燃油消耗和行驶工况数据进行划分,获得大量行驶片段。接着用主成分分析法从所有行驶片段的13个特征参数中得到了3个主成分。最后利用BP神经网络对3个主成分的得分进行燃油消耗的预测。结果表明,与一般的BP神经网络相比,采用主成分分析和神经网络相结合的燃油消耗预测模型,简化了网络结构,提高了预测精度,可用来预测城市道路行驶工况的燃油消耗。
Fuel consumptions are predicted according to the characteristic parameters of driving cycle in the paper. Firstly the collected data on 100km fuel consumption and driving cycles on typical roads are divided into a large number of driving segments. Then three principal components are obtained from 13 characteristic parameters of all driving segments by principal component analysis. Finally fuel consumptions are predicted by BP neural network based on the scores of three principal components. The results show that compared with BP neural network, the fuel consumption prediction model based on the combination of principal component analysis and neural network can better predict the fuel consumption of driving cycle on urban roads with simplified network structure and improved prediction accuracy.
出处
《汽车工程》
EI
CSCD
北大核心
2014年第6期643-647,共5页
Automotive Engineering
基金
国家自然科学基金(71071044)
合肥工业大学校博士基金(2011HGBZ1294)资助
关键词
燃油消耗预测
特征参数
BP神经网络
主成分分析
fuel consumption prediction
characteristic parameters
BP neural network
principal com-ponent analysis
