DES-KNORA Model: Developing an Intelligent System for Automatic Power Mode Adjustment in Electric Bicycles

Electric bicycles (e-bikes) have gained considerable popularity due to their environmentally friendly nature and suitability as a mode of transportation. However, they face challenges related to manual switches for power mode changes. A potential solution to this issue could be the application of machine learning to improve their performance and safety. Previous research has explored the use of machine learning for gear shifting in bicycles, but there remains a gap in the specific application of ML for automating power mode changes in electric bikes. In order to address this gap, the study proposes the design and evaluation of an intelligent system based on machine learning that can automatically adjust electric bike power modes for users. The research team utilized a Hero Lectro C7+ electric bike and designed and tested a novel ensemble model, the dynamic ensemble selection approach model (DES-KNORA), against several machine learning techniques to predict the need for power mode changes. The DES-KNORA ensemble model fit the measured data the best, with an F1-score of 0.9995 and an accuracy of 99.95%. The study’s results show that the suggested method works in real life, making the ride smoother and less annoying for riders.