论文部分内容阅读
针对冲击地压影响因素模糊、非线性等特性,选用一种混沌果蝇优化算法改进的广义回归神经网络(CFOA-GRNN),通过引入混沌扰动因子,增加网络的参数搜寻能力,建立冲击地压危险等级预测模型。模型选取煤层厚度、埋深等10个主要影响因素。以砚台煤矿的35个样本数据对模型进行学习、训练和预测,并将预测结果与传统预测模型的结果进行比较;同时采用敏感性分析法,评价各主要因素对模型精度的影响。结果表明煤层倾角影响最大,其次为卸压方式。研究表明,所构建的CFOA-GRNN模型正确率为92.8%,明显优于传统预测模型。
A fuzzy regression neural network (CFOA-GRNN) improved by Chaos Drosophila optimization algorithm is selected for the fuzzy and nonlinear characteristics of rock burst. By introducing chaotic perturbation factor and increasing the parameter search ability of the network, Hazard level prediction model. The model chooses 10 main influencing factors, such as coal seam thickness and depth. The model was studied, trained and predicted by using 35 sample data of Yantai Coal Mine. The prediction results were compared with the results of the traditional prediction model. At the same time, the sensitivity analysis method was used to evaluate the influence of the main factors on the model accuracy. The results show that coal seam dip has the greatest impact, followed by pressure relief mode. The research shows that the correct rate of CFOA-GRNN model constructed is 92.8%, which is obviously better than the traditional prediction model.