Virtual Sensing and Sensors Selection for Efficient Temperature Monitoring in Indoor Environments

Area of Science:

• Environmental monitoring
• Sensor networks
• Artificial intelligence

Background:

• Real-time indoor temperature monitoring is crucial for comfort, legal compliance, and energy efficiency.
• Deploying adequate physical sensors for uniform temperature monitoring can be challenging and costly.
• Virtual sensing offers a solution by replacing physical sensors with virtual ones, enabling monitoring of inaccessible areas and reducing deployment expenses.

Purpose of the Study:

• To develop a versatile, black-box virtual sensing framework for indoor temperature monitoring, independent of specific environmental characteristics.
• To systematically analyze and develop novel distance metrics for optimal sensor selection.
• To create an automated approach for identifying the most effective subset of sensors and predicting temperatures.

Main Methods:

• Systematic analysis of various distance metrics for sensor selection.
• Genetic programming to design a novel, combined distance metric.
• Development of a general, automated approach for subset sensor selection.
• Assessment of different prediction strategies using selected sensors and evaluation of output reliability.

Main Results:

• A novel distance metric was developed using genetic programming, outperforming individual distance metrics.
• An automated method for selecting the optimal subset of sensors was proposed.
• The proposed black-box virtual sensing framework demonstrated accurate and efficient temperature monitoring in an open office scenario.
• The framework successfully derived a virtual sensing model capable of predicting temperatures based on selected sensor data.

Conclusions:

• The developed black-box virtual sensing framework is effective for accurate and efficient indoor temperature monitoring.
• The automated sensor selection and virtual sensing model generation can be adapted to various indoor environments.
• This approach offers a cost-effective and reliable solution for temperature monitoring challenges.