中文摘要：

为了提高煤矿冲击地压预测预报的准确率,在综合考虑自然因素和开采因素的基础上,针对煤矿冲击地压系统小样本、多维度、非线性的特点,提出煤矿冲击地压预测的改进网格搜索支持向量机模型（GS-SVM）。利用该模型对四川某矿历史统计数据进行预测分析,并与启发算法优化支持向量机参数模型、神经网络模型、Fisher判别分析模型,传统网格搜索优化支持向量机模型进行比较。结果表明：改进GS-SVM模型能够对具有多维度、非线性、小样本特征的冲击地压进行很好的预测预报,与其他模型相比训练时间更短,预测精度更高,对煤矿冲击地压预测及防治具有一定的指导意义和参考价值。

英文摘要：

In order to improve the accuracy of coal mine rock burst prediction and forecast, an improved grid search and sup- port vector machine model is proposed. Tile natural factors and mining factors are considered comprehensively, making the model suitable for the charaeteristies of small sample,multi dimension and nonlinear of coal mine impact pressure system. The historical statistical data of a mine in Sichuan is used for the prediction of the model. The model and ＂ Heuristic algo- rithm＂ optimization support vector machine parameter model,neural network model, Fisher discriminant analysis model, and the model of traditional grid search optimization support vector machine model are compared. The results show- that the im- proved GS -SVM model can predict the impact ground pressure with multi dimension,nonlinear and small sample charac- teristics. Compared with other models,the improved GS-SVM model has shorter training time and higher prediction accura- cy, which has certain guiding significance and reference value for the prediction and prevention of mine rock burst.