Sustainable two-wheeler electrification: forecasting the optimal age for electric vehicle retrofitting using artificial intelligence

Retrofitting internal combustion engine (ICE) two-wheelers to electric vehicles offers a practical pathway for reducing emissions while extending the vehicle's lifetime at a lower cost than purchasing new electric models. However, the lack of objective criteria to determine when a vehicle should be converted remains a major barrier to large-scale adoption. In this work, an experimental framework is developed to identify the retrofit feasibility boundary for ageing ICE two-wheelers by jointly analysing mechanical condition, emission behaviour, and performance degradation. Four TVS XL 100 vehicles aged 4, 6, 9, and 10 years were evaluated using chassis dynamometer testing, tri-axial vibration measurements, and exhaust emission analysis. High-frequency vibration data were acquired at 10 kHz, resulting in thousands of feature observations for each vehicle condition. A Component Health Index (CHI) was formulated to quantify degradation by integrating vibration-based mechanical indicators with combustion and performance metrics. These multi-domain degradation indicators were used as a supervised learning based machine learning model that serves as the predictive retrofit feasibility component. The Deep Spider-Wasp Belief Network (DSWBN) model predicts the retrofit feasibility and is then processed through an AI-based fuzzy inference decision layer to convert probabilistic results into a clear and actionable retrofit decision. The results indicate that, for the investigated TVS XL 100 vehicles under the considered operating and maintenance conditions, a vehicle age of approximately nine years exhibited the most favourable balance between mechanical integrity, degradation severity, and retrofit feasibility. A complete hardware conversion was implemented using a 60V, 30Ah lithium-iron-phosphate battery and an 800W BLDC hub motor. Experimental results closely matched simulation outcomes. Techno-economic analysis shows annual savings of Rs. 79,396 with a payback period of approximately 4.9 months.