关键词: 矿井定位, 指纹定位, 克里金插值, 粒子群, 模拟退火

Abstract: Location information is very important for personnel management and post-disaster rescue in mine operation. Fingerprint location can significantly improve the accuracy of mine location,but the time-consuming and laborious process of fingerprint database construction hinders its large-scale application. Therefore,a Kriging interpolation algorithm (PSO-SA-Kriging) based on particle swarm optimization and simulated annealing was proposed. Only partial data of fingerprint sampling points were collected in the initial stage,and all fingerprint data sets were obtained by Kriging model interpolation based on the data of these sampling points,and the Kriging theoretical model was optimized by particle swarm optimization and simulated annealing algorithms. The constructed fingerprint database is more suitable for the actual mine environment. The algorithm takes advantage of the particle swarm convergence speed to solve the problem of rapid construction of fingerprint database. At the same time,simulated annealing is used to overcome the defect that particle swarm may fall into local optimal,which makes model fitting more accurate and interpolation results more accurate. Based on the above analysis results,by collecting fingerprint data in the mine environment,a total sampling database,a mixed database of half sampling and interpolation are established,and the nearest neighbor algorithm (KNN) is used to verify the proposed method. The results show that POS-SA-Kriging algorithm not only greatly reduces the fingerprint construction workload,but also significantly improves the location accuracy,realizing the joint optimization of the fingerprint database construction speed and target location accuracy.