基于改进FOA-SVM的冲击地压危险性等级预测

Prediction of rock burst risk grade based on improved FOA-SVM

对冲击地压危险性进行准确的预测预报对于防治冲击地压事故的发生至关重要。提出利用改进的果蝇优化算法(FOA)优化参数,建立模型实现对冲击地压危险性等级的预测。首先,利用文献提供的砚石台煤矿实测数据作为样本,选取影响冲击地压发生的十种主要因素如煤厚、埋深、倾角等,对数据进行归一化预处理和主成分分析。利用改进FOA的全局优化能力对SVM进行寻优,继而建立FOA-SVM模型;然后对23组训练样本进行训练,检验得模型误判率为0;最后将模型用于另外12组现场采集数据进行测试,并与标准FOA-SVM、PSO-SVM和GA-SVM预测结果进行比较。结果表明:改进的FOA-SVM模型适用于冲击地压危险性等级预测且预测精度较高。

The accurate prediction of rock burst risk is very important to prevent and controll rock burst accident. The model optimized by improved Fruit Fly Optimization Algorithm（ FOA） is proposed to predict the risk grade of rock burst. Firstly,the measured data of Yanshitai coal mine provided by the literature are taken as samples,ten main influencing factors of rock burst such as the coal thickness,the buried depth,as well as the dip angle are selected. The data are processed by normalization and principal component analysis. The parameters of SVM are optimized by using the global optimization ability of improved FOA,then the FOA-SVM model is established; Secondly,23 groups of training samples are trained,and the ratio of mis-discrimination is 0; Finally,the model is used to test the other 12 sets of field data and compared with the standard FOA-SVM,PSO-SVM and GA-SVM prediction results. The results show that the improved FOA-SVM model is applicable to the prediction of rock burst risk grade and the prediction accuracy is higher.