Forest Fire Prevention Early Warning Method Based on Fuzzy Bayesian Network

Abstract

In the environment of large forest, the factors causing fire are nonlinear and uncertain. If the data collected by the sensor is simply analyzed and compared, the false alarm rate will be higher. How to combine the data of several sensors for effective fire warning is a difficult point. In order to improve the accuracy of prediction, aiming at the shortcomings of traditional forest fire prevention early warning system, we propose a forest fire prevention early warning method based on fuzzy Bayesian network. Firstly, we combine the fuzzy control system and the Bayesian network in series, and pre-process the collected sensor data. The pre-processed data is sent to the previously trained Bayesian network for processing. Then the calculated open fire probability, smoldering fire probability, and no fire probability are used as input data of fuzzy control system, and fuzzy inference is performed. Finally, we de-fuzzify the results of fuzzy reasoning and get the probability of fire. Simulation results show that our method can effectively combine the data collected by multiple sensors, quickly and accurately determine fire occurrence probability, improve the accuracy of forest fire prevention warning, and reduce the false positive rate.