基于PSO-SVM的煤矿瓦斯爆炸灾害风险模式识别

Research on Pattern Recognition of Gas Explosion Disaster Risk in Coal Mines Based on PSO-SVM

为了提高煤矿瓦斯爆炸灾害风险识别能力,提出通过典型样本和模式识别(PR)模型构建识别库进行风险识别的方法。在以径向基函数(RBF)为核函数的支持向量机(SVM)模型基础上,应用粒子群优化算法(PSO)优化模型参数,建立瓦斯爆炸灾害风险识别的PSO-SVM模型;构建风险识别指标体系,确定风险模式类别,并以指标风险类别分界点为基础,提出新的数据规范方法;用典型样本训练和测试PSO-SVM模型,样本识别率为100%,表明以典型样本和PSO-SVM模型构建的识别库对瓦斯爆炸灾害风险有较强的识别能力,同时指出典型样本库应不断补充完善以增强其适应能力。

To improve the risk identification of gas explosion disaster in coal mines,an approach was made to combination of a model for gas explosion pattern recognition with typical risk samples of gas explosions.A PSO-SVM model was built for the risk identification.The PSO-SVM model was trained and tested with the typical samples,the recognition rate of the sample was 100%.