Detect defective components thanks to AI
Error codes or sensor values allow mechanics to identify defective components, but this currently requires extensive automotive knowledge and a lot of time. The large number of different vehicle types also increases the complexity of the challenge. Machine learning methods can help to automatically detect potentially defective components while driving.
How does this work? Historical vehicle data that has already been collected (such as fault codes, sensor readings and mileage) is used as training data (features) for model development. For this purpose, the machine learning method Feature Engineering is used. Feature engineering facilitates the analysis of data. It does not matter if it is an SQL database, an Excel list or any other data source. Feature engineering is about putting the given data into a form that is easier for algorithms to interpret.
Then, data sources that contain information about components that are potentially defective or have been replaced in the vehicle are linked as target variables. Example data are invoice data from workshops or repair information from mechanics on the diagnostic device.
Thanks to the integration of machine learning processes, vehicle diagnostics can be carried out more efficiently. The complexity due to different vehicle types can be better managed.