The charging infrastructure design problem with electric taxi demand prediction using convolutional LSTM. (7th October 2022)

• Record Type: Journal Article
• Authors: Hwang, Seong Wook; Lim, Sunghoon
• Is Part Of: European journal of industrial engineering. Volume 16:Number 6(2022)
• Page Start: 651
• Page End: 678
• Publication Date: 2022-10-07
• Subjects: OR in service industries -- transportation -- heuristics -- machine learning -- artificial intelligence; Industrial engineering -- Europe -- Periodicals; 658.50094
• Languages: English
• ISSNs: 1751-5254

Abstract

The authors present a charging infrastructure design problem with electric taxi demand prediction. Due to environmental concerns, electric vehicle adoption has significantly increased in the transportation sector. However, the use of electric vehicles is not highly commercialised in the taxi industry, because the immature charging network and frequent charging decrease taxi revenue efficiency. Therefore, charging infrastructure needs to be built in urban areas in consideration of operational requirements of the taxi industry. The authors first design a convolutional long short-term memory model that predicts taxi demand, along with hotspots. Then, based on the predicted taxi demand in hotspots, a mixed integer linear programming model is proposed to optimise the location of recharging stations to minimise the cost of locating stations and charging service. Also, we propose a heuristic algorithm to solve realistic and practical problems. Lastly, a case study is presented to validate the proposed research. [Submitted: 28 April 2021; Accepted: 5 September 2021]