关键词: 通风系统, 风机定位, 大规模非线性混合整数规划, 鱼群算法

Abstract: In order to solve the optimization problem of location of fans in a ventilation system of underground mine，a large-scale nonlinear mixed integer programming model is established for this problem.In the optimization model，the solving space is discretized based on actual consideration of space constraints of variables and precision of solutions.The policy of discretization is to discrete each real variable into finite values and quantize the continuous interval of each variable into finite levels.Because the dimension of the model is so enormous that traditional optimization techniques cannot find its optimal solution effectively，an improved fish-swarm algorithm is used to solve the problem.In the improved algorithm，in order to delete constraints and compress greatly the variables of the model，all variables are grouped to reduce the number of variables.An evolutionary algorithm based on the neighbor competitive evolution operator is applied so that the basic properties of global and local search are mixed together and a balance between efficiency and effectiveness is realized.The multi-agent-based self-leaning and self-adaptive variation operator，the orthogonal crossover operator，the neighbor competitive evolution operator and so on are used to improve all behaviors of artificial fish-swarm algorithm.An application result shows that the speed and reliability of the algorithm has a significant improvement in the optimization algorithm，and the solution of the model can be gained in the simple computing environment.

Key words: ventilation system, locating of fans, large-scale nonlinear mixed integer programming, fish-swarm algorithm