Link Quality Estimation for Wireless ANDON Towers Based on Deep Learning Models.

Area of Science:

• Industrial IoT and Machine Learning
• Wireless Sensor Networks

Background:

• Data reliability is crucial for industrial decision-making.
• Wireless sensor networks (WSNs) are vital for process and machine monitoring.
• ANDON towers with wireless transmission and machine learning can predict link quality (LQE).

Purpose of the Study:

• To propose a deep learning model for LQE prediction suitable for resource-limited industrial environments.
• To enable low-cost remote sensing and early failure detection in industrial settings.
• To develop a novel paradigm for ANDON towers using alarm signals and LQE classification.

Main Methods:

• Collected a novel dataset from a realistic industrial machinery scenario.
• Utilized a deep learning model optimized for limited computational resources.
• Performed extensive data analyses with methodical hyper-parameter tuning across various machine learning models.

Main Results:

• Achieved 99.3% accuracy on the test dataset.
• Identified key features like payload, distance, power, and bit error rate for LQE prediction.
• Demonstrated high performance with minimal computational resource utilization.

Conclusions:

• The proposed deep learning model offers an effective solution for LQE prediction in industrial environments.
• This approach facilitates cost-effective remote sensing and proactive problem prevention.
• The findings advance the state of the art in industrial wireless communication and machine learning applications.