Usage of IR Sensors in the HVAC Systems, Vehicle and Manufacturing Industries: A Review

Thermal sensors are extensively used in industrial applications for process monitoring, control, and measurement, providing superior visibility in low light, dark, and adverse weather conditions. They are widely employed in machine vision applications such as industrial inspection, security monitoring, and autonomous vehicle perception.

These sensors absorb radiation to change material temperature, generating an electrical output. They are user-friendly, cost-effective, and suitable for non-scanned images. Thermal sensors can be applied in diverse fields, including human detection, medical imaging, robotics, lighting control, and Heating, ventilation, and air conditioning (HVAC) systems.

This survey explores using Infrared (IR) sensors as thermal sensors in various industrial and smart applications, highlighting the types of sensors available, their common applications, and a detailed review of application-specific methods.

Materials with a temperature above 0 K emit thermal energy in the form of IR radiation, ranging from 0.7 to 1000 μm. The 0.7-to-20-micron range is commonly used for temperature measurement as IR detectors are less sensitive beyond 20 microns.

IR sensors in activity recognition techniques are being developed along with multiple sensors, particularly in smart homes and other intelligent devices. These sensors collect information about the temperature, movement, and behavior of individuals. Passive Infrared Sensors (PIR) provide privacy but cannot differentiate among humans and require preliminary information about house layout.

Thermal sensors, such as thermal imaging cameras, excel in day and night-time detection by converting the infrared light spectrum into images. They are passive, less affected by scattering, and perform well in challenging weather conditions. Their high resolution enhances compatibility with Radar and LIDAR, making them particularly effective for night-time target detection, especially for small objects and pedestrian classification. These cameras are used in autonomous vehicles and surveillance systems for robust image acquisition in various environments.

Thermal imaging uses cooled and uncooled sensors, with uncooled sensors functioning at ambient temperature and cooled sensors requiring low temperatures for high sensitivity but higher cost and lower reliability. Uncooled systems are mobile-friendly, affordable, and suitable for diverse applications, while cooled systems are economical for longer ranges.

The study presents a comparative analysis of varied detection methods such as thermal infrared network (TIN), Path Aggregation Network (PAN), Support vector machine (SVM), and convolution neural network (CNN) for different detection applications. The random forest ML algorithm coupled with the MLX90640 sensor showed 99% accuracy, making them ideal for HVAC systems and smart buildings.

A multi-modal system, integrating thermal and PIR sensors, is employed for occupancy estimation. Challenges arise when a person remains stationary for extended periods, which is tackled by various machine learning algorithms to enhance the adaptability of thermal sensors.

IR cameras are also used for contactless detection of issues in photovoltaic panels, such as cracks and hotspots. However, accurate fault location and defect classification still need to be addressed. Additional methods include image processing, feature matching, threshold methods, autonomous flying, and machine learning algorithms.

The study has showcased that selecting the suitable algorithm and strategically placing sensors is as crucial as choosing the appropriate sensor, considering the situation and desired output results for specific applications.