汽车工程学报2018，Vol.8Issue(3)：182-188,7.DOI:10.3969/j.issn.2095-1469.2018.03.04
汽车试验场在场车辆总数趋势预测
Trend Forecasting of the Total Number of Vehicles on Automobile Proving Ground
摘要
Abstract
In order to analyze the total number of vehicles on an automobile proving ground for the past 3 years,and to forecast the trend for the coming year,this paper has adopted ARIMA and Fbprophet methods for some preliminary analysis. However,it is found that these traditional statistic methods cannot identify the remote points in the data,and are not suitable for long-term forecasts. Moreover,LSTM and GRU methods are used respectively to train the total number of vehicles,and for both methods the cost function can converge quickly. And then the trained models are applied to the test data and the results show that the LSTM method achieves higher accuracy in prediction.关键词
汽车试验场/车辆总数/趋势预测/LSTM/GRUKey words
automobile proving ground/vehicle's total number/trend predict/LSTM/GRU分类
交通工程引用本文复制引用
曾敬,向华荣..汽车试验场在场车辆总数趋势预测[J].汽车工程学报,2018,8(3):182-188,7.基金项目
工信部2016年工业强基工程(0714-EMTC02-5593/20) （0714-EMTC02-5593/20）
