Application of Improved DPO-BP Algorithm in Fire Detection
In order to solve the problems of low accuracy,false alarm and underreporting of BP neural net-work in fire detection of high-rise buildings,an improved dung beetle optimization algorithm(DBO)was pro-posed to optimize BP neural network to achieve fire detection.The input of BP neural network is the concentration of CO,temperature and the concentration of smoke,and the output is open fire,smoldering fire and non-flame.Firstly,tent mapping was added to the initial population generation of dung beetle optimization algorithm to im-prove,so as to generate an initial population with uniform distribution and good diversity.Then the weight and threshold parameters of BP neural network were optimized by the improved dung beetle algorithm,and the opti-mal IDBO-BP fire prediction model was constructed.Finally,The simulation comparative experiments were conducted on the BP model,DBO-BP model,and IDBO-BP model.Simulation results show that IDBO-BP fire prediction model can detect fire faster and more accurately than BP and DPO-BP model,and the accuracy rate is improved to 98.99%,which strengthens the reliability of fire detection.
BP neural networkfire detectiondung beetle optimization algorithm
NETL
NSTL
万方数据
