Application of Self-Adaptive Neural Fuzzy Network in Early Detection of Conveyor Belt Fire

With the more usage of conveyor belts in coalmines, the possibility of fire is increasing and the loss caused by conveyor belts fire is becoming more and more. The traditional alarm system using single sensor or relatively simple algorithm processing fire signals cannot meet the need of belt fire detection in coal mine. Multi-sensor fire detection system is one of the important current developments in automatic fire detection technology. The crucial point in the detection is using more sophisticated algorithm to process fire signals. In this thesis, we introduce the principle of an algorithm based self-adaptive neural fuzzy network which has four inputs - temperature (T), rate of temperature change (¿T), dense of carbon monoxide (CO) and rate of CO dense change (¿CO). The experiment result shows that using this system can enhance the adaptive ability to environment and reinforce the ability of resisting interfere. At the same time, combining the fuzzy inference and neural network used in processing the fire signals makes the system more intelligent and decrease the false alarm rate.