GA-BP Optimization Using Hybrid Machine Learning Algorithm for Thermopile Temperature Compensation

Abstract

Thermoelectric pile, which uses non-contact infrared temperature measurement principle, is widely used in various precision temperature measuring instruments. This paper analyzes environmental temperature's influence on thermoelectric piles' measurement accuracy and proposes a environment temperature compensation based on GA-BP (Genetic Algorithm-Back Propagation) neural network. The GA algorithm makes up for the slow iterative speed and easy to fall into local optimization of BP algorithm. The experimental simulation results show that environment temperature compensation based on GA-BP can accurately correct the measurement error caused by environmental temperature and other factors.