A Systematic AI-Based Paradigm for Classifying Hybrid Electric Vehicle Spare Parts Using Their Price and Demand

Abstract

With the rapid growth of the hybrid electric vehicle (HEV) market, stakeholders must refine after-sales logistics, particularly spare parts (SPs) provisioning. Adequate forecasting of demand and pricing underpins these efforts. This paper presents a feedforward artificial neural network (FF-ANN) that, drawing on 15 predictor variables, assigns each SP to low, medium, or high categories for both demand and price. The resulting scheme enables judicious inventory management, optimized procurement, and reduced operational spending by synchronizing supply with realistic forecasts. Explainable artificial intelligence (XAI) techniques were then incorporated to enhance model interpretability by identifying key influencing factors. The results indicated that factors such as failure rate, number of cars, car age, and average total maintenance cost have a high relative impact on demand prediction outcomes, whereas factors such as part type, online price, and the new/used parts have high impacts on the price prediction model.