Fire judgment method based on intelligent optimization algorithm and evidence fusion.

To reduce the adverse effects of inevitable error and random error in the fire data obtained by multiple sensors, in this paper, we propose a fire judgment method that uses Swarm intelligence optimization techniques and Evidence fusion. First, three sensors (CO, smoke and temperature) are used to obtain fire data which are processed by the method of the interval number processing, and the distances between the fire data and the characteristic value of fire grade are calculated. The reliability coefficient N k is optimized by Swarm intelligence algorithms to complete the modification of the mass function. Then, the combination rule of interval evidence and the modified mass function are synthesized to obtain the comprehensive interval evidence. Finally, the fire grade is judged according to the decision rule. We study the usability of these techniques for fire judgment and compare the optimization performance of the important Swarm intelligence algorithms, including traditional Particle Swarm Optimization (PSO ) and its improved algorithm (IPSO ), the latest algorithms, Black Widow Optimization (BWO ) and its improved algorithm (IBWO ), Bald Eagle Search Algorithm (BES ) and its improved algorithm (IBES ). The experimental results show that the average probabilities of IBWO , IBES and IPSO for obtaining the correct fire grades are 0.96, 0.88, and 0.86, respectively, the performance of three improved algorithms in fire judgment have been greatly increased, compared to traditional D - S evidence fusion method, the increase ratios of IBWO , IBES and IPSO are 43.3%, 31.3%, 28.4%. Therefore, the D - S evidence fusion method optimized by Swarm intelligence algorithms are better than that of traditional D - S evidence method for fire detection, which provides a new idea for fire detection. [ABSTRACT FROM AUTHOR]