As a global catastrophic event, the smoke and flame produced in a short time can cause serious losses to people's lives and property. Aiming at the high false alarm rate and missing alarm rate of fire detection, a new fire detection algorithm using the BSO to optimize the ELM is proposed, which optimizes the ability of the limit learning machine to find the optimal weight and threshold, and improves the generalization ability and accuracy of the limit learning machine. The PyroSim software is used to simulate, generate sample data, train the BSOELM, and verify the superiority of the algorithm by comparing it with the ELM and the PSO-ELM.

fire detection algorithm, PyroSim, beetle swarm optimization, extreme learning machine, particle swarm algorithm