Thermal Image Learning for Air Temperature Distribution Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for learning temperature distribution in a target space rely on information from an air conditioner's blowing outlet, which is not available when there is no air conditioner, preventing accurate temperature distribution learning.
Innovation Solution
A learning model associates temperature distributions with thermal images to learn the temperature distribution without relying on air conditioner information, using thermographic cameras and temperature sensors to create a model that estimates temperatures at multiple points in the air, and identifies and reflects heat source influences.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If information from air conditioner blowing outlet is used to learn temperature distribution, then temperature distribution learning can be performed in spaces with air conditioners, but the method fails when there is no air conditioner in the target space
Solution Approach 1:
The patent introduces thermal images as an intermediary variable to bridge the gap between observable data and temperature distribution. Instead of directly using air conditioner blowing outlet information, the system uses thermal images captured by thermographic cameras as a mediator to infer temperature distributions, enabling the method to work in spaces without air conditioners while maintaining learning accuracy through the intermediary thermal data
Solution Approach 2:
The patent creates a copying approach by using thermal images as a substitute representation for direct temperature measurements. The thermal image serves as a copy or proxy that contains information about temperature distribution, allowing the system to learn temperature characteristics without requiring physical temperature sensors or air conditioner data, thus expanding applicability to various spaces
2Measurement precision
If multiple temperature sensors are installed to accurately estimate temperature distribution, then estimation accuracy improves, but installation costs and device complexity increase
Solution Approach 1:
The patent replaces the mechanical system of multiple physical temperature sensors with an optical/thermal imaging system. Instead of using numerous sensors that physically contact or proximity-detect temperatures, the system uses thermographic cameras to capture thermal radiation patterns, substituting a non-contact optical measurement approach for the mechanical sensor array, thereby reducing device complexity while maintaining or improving measurement precision
Solution Approach 2:
The patent uses thermal images as a copy representation of temperature distribution rather than relying on direct sensor measurements. The thermal image captures temperature information across the entire space as a visual pattern, allowing the system to estimate temperature distribution at multiple points without installing physical sensors at each location, thus reducing the number of components needed
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables accurate temperature distribution learning and estimation in spaces without air conditioners, reducing the need for numerous sensors and allowing for heat source influence reflection, thereby enhancing estimation accuracy and reducing installation costs.
Implementation Method 1
there is a technique of learning the temperature distribution of a target space using thermal images of the target space
Data Source
AI summary
A temperature distribution learning apparatus learns a temperature distribution including a temperature at least a first point in air of a target space. The temperature distribution learning apparatus includes a learning model. The learning model learns the temperature distribution, which is an objective variable, and a thermal image, which is an explanatory variable and is related to the target space, in association with each other. The learning model learns based training data including the temperature distribution and the thermal image.