As the era of autonomous driving approaches, the vehicle's ability to detect abnormalities on its own has become essential
With the rapid growth of the autonomous vehicle and car-sharing markets, there is an increasing demand for self-diagnostic technology that allows vehicles to diagnose their own condition without human intervention. Among these, the wheel bearing is a core component that directly receives road impacts and loads during driving.
However, in actual driving environments, accurate abnormality detection was not easy due to the complex mixture of speed changes and external noise. The key challenge was to create a highly reliable diagnostic system using only low-cost commercial sensors without expensive dedicated sensors.
Removing Noise to Isolate Critical Signals for AI Diagnostics
That Work Under Any Driving Condition
First, data was collected from vehicles operating in real-world driving environments to analyze the characteristics of external noise. Noise filtering and stable signal extraction techniques were developed to establish a foundation for obtaining consistent signals regardless of driving conditions.
Based on these signals, both physics-based and data-driven features were extracted to quantitatively evaluate wheel bearing health. Finally, an AI-based classification algorithm was applied to automatically determine the presence of faults, completing the autonomous diagnostic framework.
Autonomous Diagnostic System That Performs Reliably Under
All Driving Conditions
Reliable diagnostics are possible even with low-cost sensors. Wheel bearing faults can be detected using only commercial MEMS vibration sensors, without the need for expensive dedicated equipment. By implementing a PHM system at a cost suitable for mass-produced vehicles, the project established a foundation that combines both practicality and reliability.
Consistent diagnostics are maintained even in complex driving environments. Thanks to noise filtering and stable signal extraction techniques, reliable diagnostic results can be achieved even when vehicle speed or road surface conditions change. The algorithm was validated on actual roads rather than only in laboratory environments, confirming its potential for real-world deployment.
This project established the technological foundation for self-diagnosing vehicles. Going beyond simple fault detection, it laid the groundwork for an autonomous diagnostic system that enables vehicles to assess the condition of their own components and detect abnormalities. It represents a meaningful step toward the era of autonomous driving.
...
...
...
...
As the era of autonomous driving approaches, the vehicle's ability to detect abnormalities on its own has become essential
With the rapid growth of the autonomous vehicle and car-sharing markets, there is an increasing demand for self-diagnostic technology that allows vehicles to diagnose their own condition without human intervention. Among these, the wheel bearing is a core component that directly receives road impacts and loads during driving.
However, in actual driving environments, accurate abnormality detection was not easy due to the complex mixture of speed changes and external noise. The key challenge was to create a highly reliable diagnostic system using only low-cost commercial sensors without expensive dedicated sensors.
Removing Noise to Isolate Critical Signals for AI Diagnostics
That Work Under Any Driving Condition
First, data was collected from vehicles operating in real-world driving environments to analyze the characteristics of external noise. Noise filtering and stable signal extraction techniques were developed to establish a foundation for obtaining consistent signals regardless of driving conditions.
Based on these signals, both physics-based and data-driven features were extracted to quantitatively evaluate wheel bearing health. Finally, an AI-based classification algorithm was applied to automatically determine the presence of faults, completing the autonomous diagnostic framework.
Autonomous Diagnostic System That Performs Reliably Under
All Driving Conditions
Reliable diagnostics are possible even with low-cost sensors. Wheel bearing faults can be detected using only commercial MEMS vibration sensors, without the need for expensive dedicated equipment. By implementing a PHM system at a cost suitable for mass-produced vehicles, the project established a foundation that combines both practicality and reliability.
Consistent diagnostics are maintained even in complex driving environments. Thanks to noise filtering and stable signal extraction techniques, reliable diagnostic results can be achieved even when vehicle speed or road surface conditions change. The algorithm was validated on actual roads rather than only in laboratory environments, confirming its potential for real-world deployment.
This project established the technological foundation for self-diagnosing vehicles. Going beyond simple fault detection, it laid the groundwork for an autonomous diagnostic system that enables vehicles to assess the condition of their own components and detect abnormalities. It represents a meaningful step toward the era of autonomous driving.