基于GA-BP的煤巷围岩稳定性分类与应用

Surrounding Rock Stability Classification and Application of Coal Roadway Based on GA-BP

为了对煤巷围岩稳定性进行科学分类,统计24篇分类文献,建立SPSS分类指标数据库,进行频数分析,研究指标的频数和分布特征,得出重点分类指标,同时考虑指标选取原则和山西焦煤实际情况,确定了11个分类指标。联合应用遗传算法和BP神经网络,建立GA-BP煤巷围岩稳定性分类模型,通过GA全局启发式搜索构建网络拓扑结构,通过GA全局寻优和BP局部优化确定最优权值阈值,应用MATLAB对该模型进行编程。设计含80条山西焦煤煤巷的训练样本,使模型学习并获得分类能力,并应用于山西焦煤20条煤巷。结果表明:GA-BP分类模型准确率为95%,具有较高非线性映射精度,适用于对山西焦煤煤巷进行稳定性分类。

In order to get scientific results of surrounding rock stability classification, referring to statistics of 24 classification related papers, the SPSS database of classification index is set up to analyze＇frequency as well as study the index frequency and distribution characteristics, then the important classification indexes are obtained. Taking the selection principles and the actual situations of Shanxi Coking Coal Group into consider, the eleven classification indexes are determined. By combined use of genetic algorithm and BP neural network, the GA - BP based coal roadway surrounding rock stability classification model is established. The network topology of the model is constructed by GA global heuristic search, and the optimal weight and threshold of the model is determined by GA global opti- mization and BP local optimization. The model is implemented by MATLAB programming. Eighty coal roadways are chosen as training samples to train the classification ability of the model, then the GA - BP based classification model is applied to twenty coal roadways. The results show that the model accuracy is 95%, with its high nonlinear mapping accuracy, it is suitable for the stability classification of coal roadways in Shanxi Coking Coal Group.