Energy consumption analysis of metropolitan logistics vehicles based on an ensemble K-means long short-term memory model

Abstract

In recent years, creating a green and low-carbon sustainable development has received extensive attention, prompting considerable research into reducing pollution emissions in the transportation sector. This paper analyzes the energy consumption patterns of logistics vehicles on Beijing’s Sixth Ring Road. Firstly, driving segments are categorized based on variations in vehicle speed, followed by the application of the [Formula: see text]-means algorithm for segment clustering, resulting in the identification of three distinct driving states and the construction of corresponding driving cycles. It is observed that the driving states have high correlations with different road grades. Subsequent analysis reveals that speed, torque, and engine speed are the primary factors influencing energy consumption of logistic vehicles. Furthermore, energy consumption prediction models using the long short-term memory algorithm for the identified driving states on various road types are built leveraging historical data, i.e. vehicle speed, motor torque, and engine speed. Finally, the analysis highlights a notable increase in 100 km energy consumption for logistics trucks on branch roads with complex road conditions. This study contributes to the effective management of energy consumption in medium and large trucks.