Case
Car engine fault sound diagnosis using AI
Experienced mechanics can identify very specific faults in a car, just from the sounds of an idling engine; it is the accuracy of sound analyses that matters. According to the study conducted by Hyundai Motor Group, the comparative trial of AI versus experts demonstrated that using sound analysis only 8.6% of experts made the correct diagnosis, while AI accuracy was 87.6%.


Problem
It is crucial to assess the condition of an engine while performing comprehensive vehicle valuations for insurance or used car sales. This task is hardly possible to complete for a manager from an insurance or motor trade company if they do not have specific knowledge and engine diagnostic equipment.
Today, detection of the fault and its cause requires hours or even days of skilled work by car mechanics.
Solution
An AI module for a mobile app uses real-time sound samples of an idling engine and performs an initial diagnosis of it. The module detects if there are any engine faults and identifies their nature with an accuracy exceeding 90%.
How it works
The trained AI module integrated into a manager’s automated workstation backend receives engine sound in real-time, detects an anomaly and its nature, and reports it.
The AI module can work as a stand-alone application, or be integrated into other comprehensive vehicle valuation systems.

Benefits
The simple and fast tool allows any employee without specific knowledge that car mechanics have to perform an initial diagnosis of an engine with an accuracy exceeding 90%.
The module reduces the vehicle valuation time and cost.
The module prevents losses caused by poor vehicle valuation.
The module is one of the filters to prevent fraudulent vehicle valuations.

