Mine fires have always been a serious challenge to mining safety, and the existence of hidden fire sources in mine fire areas is one of the key factors causing coal seam spontaneous combustion. Hidden fire source refers to the potential fire source that cannot be easily detected by traditional detection means in the mine. They gradually accumulate heat in the complex environment of the mine. If not detected and dealt with in time, it can lead to a serious mine fire. However, traditional fire detection methods are difficult to detect hidden fire sources in the early stages, and in severe cases, can lead to catastrophic mine fires. Therefore, this study proposes a combined firefly optimisation algorithm and particle swarm optimisation algorithm for infra-red remote sensing inversion of hidden fire sources in mine fire areas, and verifies it. The results indicated that the inversion model could accurately obtain the location and intensity information of hidden fire sources. The joint algorithm exhibited higher accuracy and faster convergence speed during the training process, with an accuracy of 97.64% and an average temperature error of only 0.73°C. The average accuracy of the inversion model optimised by the joint algorithm improved to 99.48%, the average calculation time reduced to 7.25 seconds, and the efficiency has been improved by 51.92%. This study innovatively combines infra-red remote sensing inversion technology with intelligent algorithms to improve the efficiency and accuracy of detecting hidden fire sources in mine fire areas, which helps to reduce safety risks and economic losses caused by mine fires.
Read full abstract