摘要
为加快电动汽车行业低碳发展进程,调整交通运输领域能源组成结构,对于电动汽车能量消耗方面的研究成为当下重点。文中分析了天气因素、社会因素以及路网线路特性等因素对道路行车速度的影响,构建基于平均速度预测的电动汽车能耗模型;提出考虑样本相似度的长短期记忆神经网络,对电动汽车平均速度进行预测,计算汽车行驶能耗,结合空调能耗得出单位里程电动汽车总耗电量;最后,基于杭州市交通路网进行算例分析,结果表明,对比传统LSTM神经网络与BP神经网络,改进的LSTM神经网络预测精度更高、泛化能力更强。
In order to accelerate the low-carbon development of the electric vehicle industry and adjust the energy composition of the transportation sector,research on the energy consumption of electric vehicles has become the current focus.This paper firstly analyzes the impact of weather factors,social factors,and road network line characteristics on road traffic speeds,and constructs an electric vehicle energy consumption model based on average speed prediction;secondly,a long-term and short-term memory neural network considering sample similarity is proposed in this paper.The average speed of the car is predicted,the energy consumption of the car is calculated,and the total power consumption of the electric vehicle per unit mileage is obtained by combining the energy consumption of the air-conditioning.Finally,based on the example analysis of the Hangzhou traffic road network,the results show that,compared with the traditional LSTM neural network and the BP neural network,the improved LSTM neural network has higher prediction accuracy and stronger generalization ability.
作者
程江洲
余子容
程杉
阮曾成
郭思涵
Cheng Jiangzhou;Yu Zirong;Cheng Shan;Ruan Zengcheng;Guo Sihan(School of Electrical Engineering and New Energy,Three Gorges University,Yichang 443002,Hubei,China)
出处
《电测与仪表》
北大核心
2020年第20期90-97,共8页
Electrical Measurement & Instrumentation
基金
国家自然科学基金资助项目(51607105)。
关键词
电动汽车
能耗预测
城市路网
S-LSTM神经网络
electric vehicles
energy consumption prediction
urban road network
S-LSTM neural network
作者简介
程江洲(1979-),男,副教授,硕士研究生导师,主要研究方向为电力设备状态在线检测、智能电网等。Email:chengjiangzhou@ctgu.edu.cn;余子容(1996-),女,硕士研究生,研究方向为电动汽车规划与运行。Email:394036934@qq.com;程杉(1981-),男,博士,副教授,主要研究方向为新能源微电网运行与控制。Email:hpucquyzu@ctgu.edu.cn;阮曾成(1996-),女,硕士研究生,研究方向为智能配电网。Email:404960316@qq.com;郭思涵(1996-),女,硕士研究生,研究方向为配电网计量大数据分析。Email:2653534429@qq.com。
