Multivariable sales prediction for filling stations via GA improved BiLSTM

作     者：Shi-Yuan Pan Qi Liao Yong-Tu Liang 

作者机构：Beijing Key Laboratory of Urban Oil and Gas Distribution TechnologyChina University of PetroleumBeijing102249China 

出 版 物：《Petroleum Science》 (石油科学（英文版）)

年 卷 期：2022年第19卷第5期

页      面：2483-2496页

核心收录：

学科分类：12[管理学] 0202[经济学-应用经济学] 02[经济学] 1201[管理学-管理科学与工程(可授管理学、工学学位)] 020205[经济学-产业经济学] 081104[工学-模式识别与智能系统] 08[工学] 0835[工学-软件工程] 0811[工学-控制科学与工程] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：partially supported by the National Natural Science Foundation of China(51874325) Science Foundation of China University of Petroleum,Beijing(2462021BJRC009) 

主　　题：Refined oil Multivariable prediction BiLSTM Genetic algorithm Future influence 

摘      要：Accurate sales prediction in filling stations is the basis to fill in the refined oil in time and avoid the outof-stock as much as *** the defect of great“lagin the general time series model,this paper summarizes the multiple factors that influence the oil sales and develops a multivariable long short-term memory(LSTM)neural network,with the hyper-parameters being improved by the genetic algorithm(GA).To further improve the prediction accuracy,the proposed LSTM neural network is generalized to bidirectional LSTM(Bi LSTM),in which the impact of future factors on present sales can be taken into account by backward ***,different LSTM structures and genetic algorithm parameters are tested to discuss their impact on prediction *** demonstrated that genetic algorithm improved Bi LSTM model is superior to extreme gradient boosting,ARIMA,and artificial neural network,having the highest accuracy of 89%.