Temperature Compensation Modeling of Infrared Methane Detector Based on BP and RBF Neural Network

摘要

The methane is a colorless and explosive gas that has no smell, which usually exists in the coalmine beyond human awareness. Therefore methane detection is of vital importance to secure production. IR absorption spectrum analysis is an effective and original approach to gas detection, especially suitable for quick and continuous measurement of gas concentration. The biggest advantage of the infrared methane detector over the traditional catalytic beads is free of poisoning. It can provide a promising alternative to methane detection. Firstly, the paper describes the principle of IR methane detection and the influence of temperature variation. And then, the neural network modeling of temperature compensation, the architecture of the BP and RBF network, the training algorithms and the results are discussed in detail. Temperature compensation models based on BP and RBF neural networks are proposed, which provide a better alternative to the error rectification.