Research And Development Of Multi Information Fusion Fire Early Warning Detector

Fire is a kind of devastating disaster.Human factors and natural factors are very likely to cause fire.With the development of social economy,there are more and more industrial areas,commercial areas,residential areas and other public places,bring about more extensive ues of power consumption.In addition,the existence of some inflammable and explosive objects also makes the probability of fire greatly increased,which will give human wealth Production and life security pose a great threat.Therefore,how to improve the overall performance of the fire detection system and realize the early warning and accurate alarm of the fire has always been a research hot-spot issues.The traditional single type fire alarm only selects one parameter to detect through a single threshold.Due to the complexity and variability of the fire,this detection method can not make accurate early warning,there is a certain rate of false alarm and omission alarm.The multi-information fusion technology can consider multiple different parameters at the same time,and give different early warning results for different environments,which can be effective to improve the accuracy of fire warning as well as reduce the false alarm rate and omission alarm rate Therefore,it is of great significance to develop and research the fire early warning detector based on multi information fusion.In this dissertation,the necessary conditions and products of fire occurrence are analyzed and studied.Smoke concentration,temperature and CO concentration are selected as three detection parameters for multi-point detection data fusion based on Bayes algorithm,and risk coefficient fusion is carried out to improve the accuracy of detection and reduce the false alarm rate and omission alarm rate.First of all,this dissertation studies the above three kinds of single fire early warning detectors: smoke,temperature and CO gas: the photoelectric smoke detector is improved in the circuit design,and the anti-interference ability and detection sensitivity of the detector are increased by using the modulation demodulation and differential amplification of the signal;for the temperature detector,based on MLX90614 infrared temperature sensor to detect the regional temperature,the method of temperature detection based on computer digital image is used to detect the temperature rise of specific position in the space;for CO gas sensor,through the analysis of response characteristics of various gas sensors,TGS series sensors are selected to detect the CO gas concentration in the area.Then,this dissertation studies the multi-point detection data fusion algorithm based on Bayes algorithm and sets up multiple measuring points,each measuring point has the detection data of the above three parameters and their detected risk factors.Firstly,the effective measuring points are selected by the way of confidence distance,then the three-dimensional view of risk factors of each measuring point is calculated and displayed by Lab VIEW software,and finally the risk coefficient of each measuring point is fused by Bayes algorithmAt last,this dissertation tests the performance of the above single fire early warning detector,and uses the way of data fusion to carry out the fire early warning experiment.The experiment shows that the sensitivity of the sensor detection is improved,and the early warning can be carried out.This method can well combine the advantages of each sensor,and carry out fire danger warning through multiple detection points,which significantly improves the false alarm rate and omission alarm rate of the fire detection system.