BP Neural Network Modeling of Infrared Methane Detector for Temperature Compensation

摘要

In recent years, gas explosion has been a frequent accident in the coalmine. This results in great demand for novel and high quality methane detector. IR spectrum analysis, being an effective original approach in the measurement of concentration of gases, is the ideal solution that meets the requirement. The article focuses on the solution of the mathematical model of infrared methane detector with the temperature influence taken into consideration. Through deep analysis, backpropagation is applied to the real-time rectification model. In the article, the structure, the learning method and the generalization issue of the IR methane detector model are discussed in detail. The result of the modeling experiment is also given. This shows, with the BP neural network model applied, the error of the detector in respect of temperature deviation can be rectified in real-time.