Intelligent Industrial Cleaning: A Multi-Sensor Approach Utilising Machine Learning-Based Regression

Area of Science:

• Food Science and Technology
• Sensor Technology
• Machine Learning Applications

Background:

• Effective cleaning of food production equipment is critical for food safety.
• Current cleaning processes are resource-intensive, requiring significant water, energy, and chemicals.
• There is a need for advanced monitoring technologies to optimize equipment cleaning efficiency.

Purpose of the Study:

• To develop and evaluate innovative sensor-based methods for monitoring the removal of food fouling.
• To utilize signal and image processing with machine learning to predict residual fouling.
• To assess the effectiveness of optical and ultrasonic sensors in real-time cleaning assessment.

Main Methods:

• Utilized optical and ultrasonic sensors to monitor fouling removal from a benchtop rig.
• Developed tailored signal and image processing techniques for cleaning monitoring.
• Implemented a neural network regression model to predict the quantity of remaining fouling.

Main Results:

• Investigated the removal mechanisms of three distinct food fouling materials.
• Achieved high prediction accuracies for remaining fouling: 98% for area and 97% for volume.
• Demonstrated the capability of sensors and machine learning to accurately track cleaning progress.

Conclusions:

• Sensors combined with machine learning provide an effective solution for monitoring food equipment cleaning.
• This approach can lead to optimized resource utilization (water, energy, chemicals) in food processing.
• The developed technology enhances the safety and efficiency of food production environments.