Tire Condition Monitoring Using Transfer Learning-Based Deep Neural Network Approach.

Area of Science:

• Automotive Engineering
• Sensor Technology
• Machine Learning

Background:

• Tire condition monitoring is crucial for vehicle safety and economic efficiency.
• Existing Tire Condition Monitoring Systems (TCMS) are often costly, limiting widespread adoption.
• Advancements in sensor technology and machine learning offer new possibilities for cost-effective monitoring.

Purpose of the Study:

• To present a novel deep learning approach for instantaneous vehicle tire condition monitoring.
• To develop a cost-effective system utilizing readily available sensors and advanced algorithms.
• To assess the feasibility of using vibration signals for detecting tire inflation pressure changes.

Main Methods:

• Acquisition of tire vibration signals using a low-cost Microelectromechanical System (MEMS) accelerometer.
• Experimentation under various tire inflation pressure conditions.
• Application of a deep learning model for analyzing vibration data and monitoring tire condition.

Main Results:

• Demonstrated the effectiveness of using vibration signals for tire condition assessment.
• Successfully implemented a deep learning model capable of instantaneous monitoring.
• The proposed method offers a potentially low-cost alternative to traditional TCMS.

Conclusions:

• Deep learning, combined with MEMS accelerometers, provides a viable solution for real-time tire condition monitoring.
• This approach can significantly enhance vehicle safety and operational economy.
• Future research can explore further optimization and integration into standard vehicle systems.