Real time implementation of predictive maintenance for service center automation using Automobile Life Extender (ALE) based on Machine Learning

Automobile Life Extender (ALE) consists of an on-board function, machine learning model operating via cloud computing and a smart phone app. The on-board function will receive signals like engine rpm, throttle position, Brake pedal position, hydraulic pressure etc from vehicles Ecu’s. Based on this data onboard ALE module will calculate the engine load, brake circuit load etc . and sends it to the predictive maintenance model via on board IoT system. The predictive maintenance model will have historical data about the type of engine, brake system and their performance curve data acquired based on test done by its OEM. Here machine learning models play a crucial role in analyzing the vehicle data dynamically, identify drive patterns from it and predict the need for maintenance of a part or a system. A hybrid approach of training models based on Supervised and Unsupervised is incorporated, which will work with certain amount of labeled data along with a set of collected unlabeled data from the vehicle, creating an active learning strategy to make most out of the current available data. Amazon sage maker can handle training the ML models and it can be fed into Amazon Bedrock cloud agent as a customized model. The Amazon Bedrock code interpretation feature can assist in visualizing the Machine learning model. The model will be able to predict repair or replacement, or maintenance is needed for it. The results from ML model will be displayed to the driver via smartphone app. The service center can get the results from cloud to do diagnosis even before the vehicle reaches service station and allocate servicing slots for that vehicle based on their current workload, spare parts availability, vehicle owner’s suitable time.