基于SSA-BP神经网络的岩爆烈度等级预测

随着深部开采战略在我国的发展，岩爆愈加成为我国资源开采时必须面对的地质灾害之一。为提高传统误差反向传播（Back Propagation，BP）神经网络模型进行岩爆预测的准确性与有效性，采用麻雀搜索算法（Sparrow Search Algorithm，SSA）优化传统BP神经网络，提出一种基于麻雀搜索算法优化BP神经网络的岩爆预测模型(SSA-BP模型)。在考虑岩爆产生的内外因基础上，选取相关岩爆预测指标，利用国内外100例已有工程岩爆数据建立SSA-BP模型，并与传统BP模型、粒子群算法（Particle Swarm Optimization, PSO）优化支持向量机（Support Vector Machines, SVM）模型对比。结果表明：SSA-BP预测模型的有效性和准确度皆高于传统BP模型和PSO-SVM模型，同时SSA-BP模型训练集的均方误差(Mean Square Error，MSE)为0.081，比传统BP模型（0.25）降低67.7%，可为类似工程的岩爆预测提供科学依据。

Prediction of rock burst intensity based on SSA-BP neural network

With the development of deep mining strategy in China, rock burst has become one of the geological disasters that must be faced when exploiting resources in China. In order to improve the accuracy of the traditional error back propagation (BP) neural network model for rock burst prediction, this paper uses Sparrow Search Algorithm (SSA) to optimize the traditional BP neural network and proposes a rock burst prediction model (SSA-BP model) based on Sparrow search algorithm to optimize BP neural network. Based on the consideration of the internal and external factors of rock bursts, the relevant rock burst prediction indicators were selected, and the SSA-BP model was established by using the data of 100 existing engineering rock bursts at home and abroad. By comparing the effectiveness, prediction accuracy, and mean squared error with traditional BP models and the PSO-SVM model. The results show that the effectiveness and accuracy of the SSA-BP prediction model are higher than those of the traditional BP prediction model and the PSO-SVM model. At the same time, the mean square error (MSE) of the SSA-BP prediction model is 0.081, which is 67.7% lower than the traditional BP model compared with 0.25 of the traditional BP model, which can provide a scientific basis for rockburst prediction of similar projects.