Wang Wentong, Zhang Qianjun, Guo Sha, Li Zhixing, Li Zhiguo, Liu Chuanju
School of Environment and Resource, Southwest University of Science and Technology, Mianyang, 621010, China.
Heliyon. 2024 Jul 4;10(13):e34141. doi: 10.1016/j.heliyon.2024.e34141. eCollection 2024 Jul 15.
China is rich in mineral resources, and problems of goaf formed in the process of resource exploitation are serious obstacle to the development of China's economic, so it is of great significance for the assessment and management of goafs. This paper introduces emerging dung beetle optimizer (DBO) algorithm and establishes DBO-BP (back-propagation) model, at the same time, it is compared with a series of heuristic algorithms coupled with BP neural network models: PSO (particle swarm optimization) - BP model, WOA (whale optimization algorithm) - BP model, and SSA (sparrow search algorithm) - BP model. Then they are applied to evaluate the hazard of goafs, the result shows that the DBO-BP model gets the highest train set accuracy, which is at least 2.7 % higher than other models, while the DBO-BP model obtains the highest test set accuracy, meanwhile its effectiveness and stability have also been proven. Finally we apply the established DBO-BP model to evaluate the hazard of the tungsten mine goaf of Yaogangshan in Hunan Province, and its excellent practicability was confirmed. This paper may provide a reference for the solution of nonlinear engineering problems.
中国矿产资源丰富，资源开采过程中形成的采空区问题严重阻碍了中国经济的发展，因此采空区的评估与治理具有重要意义。本文介绍了新兴的蜣螂优化器（DBO）算法并建立了DBO-BP（反向传播）模型，同时将其与一系列启发式算法结合BP神经网络模型进行比较：粒子群优化（PSO）-BP模型、鲸鱼优化算法（WOA）-BP模型和麻雀搜索算法（SSA）-BP模型。然后将它们应用于采空区危险性评价，结果表明DBO-BP模型的训练集准确率最高，比其他模型至少高2.7%，同时DBO-BP模型的测试集准确率也最高，其有效性和稳定性也得到了验证。最后将建立的DBO-BP模型应用于湖南省瑶岗仙钨矿采空区危险性评价，证实了其良好的实用性。本文可为解决非线性工程问题提供参考。
