基于逻辑回归感知机结构优化蚁群算法的路径规划研究  

Research on path planning based on logistic regression perceptron structure optimized by ant colony algorithm

作　　者：郭福雁 张钰奇 王悦 秦政 GUO Fuyan;ZHANG Yuqi;WANG Yue;QIN Zheng(School of Control and Mechanical Engineering,Tianjin Chengjian University,Tianjin 300384,China)

机构地区：[1]天津城建大学控制与机械工程学院,天津300384

出　　处：《火灾科学(中英文)》2024年第4期252-262,共11页Fire Safety Science

基　　金：天津市企业科技特派员项目(19JCTPJC48900)。

摘　　要：公共建筑的空间布局和复杂性给路径规划带来了挑战,因为通道布局复杂,且存在障碍物,使得全局搜索更加困难,并可能导致算法陷入局部最优解,所选路线不平滑。提出了一种基于逻辑回归叠加感知机结构(LR-MLP)的优化蚁群算法路径规划模型,该模型使用逻辑回归模型对每条路径进行差异化处理,结合神经网络感知机结构生成初始全局优化路径。在此基础上,使用蚁群(ACO)算法进行迭代和搜索,以减少计算量和信息素更新的频率,进一步提高算法的适应能力,获得全局最优解。对比ACO和LR-MLP-ACO两种模型的仿真结果,LR-MLP-ACO模型的路径长度减少了12.8%、转折点数量减少了56%、执行时间加快了14.8%,这些数据有力地证明了该优化模型在求解问题时的有效性。The spatial layout and complexity of public buildings bring challenges to path planning because the channel layout is complex and there are obstacles,which makes global search more difficult and may cause the algorithm to fall into local optimal solutions and the selected route is not smooth.Therefore,this paper proposes an optimized ant colony algorithm path planning model based on logistic regression stack perceptron structure(LR-MLP).The model uses a logistic regression model to differentiate each path and combines the neural network perceptron structure to generate the initial global optimization path.On this basis,the ant colony(ACO) algorithm is used for iteration and search,in order to reduce the amount of computation and the frequency of pheromone update,further improve the adaptability of the algorithm and obtain the global optimal solution.Comparing the simulation results of ACO and LR-MLP-ACO models,the LR-MLP-ACO model reduces the path length by 12.8% and the number of turning points by 56%,and speeds up the execution time by 14.8%.These data strongly prove the effectiveness of this optimization model in solving problems.

关 键 词：路径规划 逻辑回归 感知机结构网络 蚁群优化算法 

分 类 号：X932[环境科学与工程—安全科学]