基于SA-PSO-BP神经网络的煤层底板破坏深度预测  

Failure Depth Prediction of Coal Seam Floor Based on SA-PSO-BP Neural Network

作　　者：李刚[1] 赵艺鸣 杨庆贺 才天 邹军鹏 Li Gang;Zhao Yiming;Yang Qinghe;Cai Tian;Zou Junpeng(College of Mining,Liaoning Technical University,Fuxin,Liaoning 123000,P.R.China)

机构地区：[1]辽宁工程技术大学矿业学院,辽宁阜新123000

出　　处：《地下空间与工程学报》2025年第1期293-299,共7页Chinese Journal of Underground Space and Engineering

基　　金：国家自然科学基金(52174077)。

摘　　要：研究煤层底板破坏深度的准确预测对保证带压开采条件下煤矿的安全生产具有重要意义。针对传统BP神经网络预测底板破坏深度存在误差较大、容易陷入局部最优解、收敛速度慢等问题,提出了一种新的SA-PSO-BP网络模型。该模型以煤层倾角、开采深度、煤层开采厚度、工作面斜长作为评判指标,先利用粒子群优化算法(PSO)改进BP神经网络寻优过程、再引入模拟退火算法(SA)避免PSO算法陷入局部最优解,选取92组现场实测数据样本,对优化后的模型进行训练和预测。结果表明:SA-PSO-BP网络模型的拟合优度达到0.9835,比BP神经网络提高了0.2882;均方根误差达到1.3190,比BP神经网络减小了3.8641;平均绝对百分比误差达到5.4423,比BP神经网络减小了14.93%。构建的SA-PSO-BP网络模型具有可行性,为底板破坏深度的预测提供了一种合理的方法。It is of great significance to study the accurate prediction of the failure depth of coal seam floor to ensure the safety production of coal mine under the condition of mining under pressure.A new SA-PSO-BP neural network model is proposed to solve the problems of large error,easy to fall into local optimal solution and slow convergence speed in the traditional BP neural network prediction of floor failure depth.The model takes the coal seam dip angle,mining depth,coal seam mining thickness and the inclined length of working face as the evaluation indexes,in order to avoid PSO falling into local optimal solution,a simulated annealing algorithm(SA)was introduced to improve the optimization process of BP neural network by using Particle swarm optimization(PSO),the optimized model is trained and predicted.The results show that the goodness of fit of SA-PSO-BP neural network model is 0.9835,which is 0.2882 higher than that of BP neural network,and the root mean square error is 1.3190,which is 3.8641 lower than that of BP neural network.The average absolute percentage error is 5.4423,which is 14.93%less than that of BP neural network.The SA-PSO-BP network model is feasible,and it provides a reasonable method for the prediction of floor failure depth.

关 键 词：带压开采 底板破坏深度 神经网络预测 SA-PSO-BP神经网络 

分 类 号：TD745[矿业工程—矿井通风与安全]