Infrared Sensor Temperature Inference Without Calibration Tables
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Infrared sensors often generate inaccurate temperature measurements due to imperfect calibration tables or minor defects in the IR detection module, leading to errors in 2D temperature matrices.
Innovation Solution
A computing device uses a neural network to bypass calibration data by generating temperature measurements directly from raw sensor data, utilizing a predictive model trained on a neural network to infer accurate temperature values.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If calibration tables are used to convert raw sensor data into temperature measurements, then the conversion process is simple and fast, but measurement precision deteriorates due to imperfect calibration adaptation to specific sensors
Solution Approach 1:
A neural network is introduced as an intermediary component between the raw sensor data and the final temperature measurements. The neural network processes the raw sensor data through learned transformations to produce accurate temperature readings, mediating the conversion process to achieve both precision and adaptability to specific sensor characteristics
Solution Approach 2:
The system changes the parameters of the conversion process by using a neural network with learnable weights and biases instead of fixed calibration table values. These parameters are trained on dataset 200 to optimize temperature measurement accuracy for specific sensor instances, allowing adaptation to minor manufacturing variations
2Ease of manufacture
If generic calibration tables are used for all IR cameras of a given model, then manufacturing cost and calibration time are reduced, but measurement precision worsens due to sensor variations
Solution Approach 1:
The neural network is pre-trained offline on dataset 200 containing raw sensor data and reference temperature measurements from multiple sensors of the same model. This preliminary training action captures the characteristics of sensor variations across the production batch, enabling the network to compensate for individual sensor differences without requiring individual calibration procedures
Solution Approach 2:
Instead of copying generic calibration tables to all sensors, the system uses a single trained neural network model that learns and copies the optimal transformation patterns from training data. The neural network captures the essence of sensor variations and applies learned corrections, effectively copying adaptive behavior rather than static calibration values
Data Source
AI summary
Method and computing device using a neural network to bypass calibration data of an infrared sensor. A predictive model generated by a neural network training engine is stored by the computing device. The computing device determines a two-dimensional (2D) matrix of raw sensor data. Each raw sensor datum is representative of heat energy collected by the infrared sensor. The computing device executes a neural network inference engine. The neural network inference engine implements the neural network using the predictive model for generating outputs based on inputs. The inputs comprise the 2D matrix of raw sensor data. The outputs comprise a 2D matrix of inferred temperatures. A method for training a neural network to bypass calibration data of an infrared sensor is also provided.