粒子群支持向量机在煤矿顶板故障诊断中的应用

The Application of PSO Support Vector Machine in Fault Diagnosis of Coal Mine Roof

利用粒子群算法和遗传算法分别对支持向量机的惩罚因子C和核参数σ进行优化。仿真结果表明,粒子群算法具有全局寻优、运算量小和操作简便等特点,其搜索最优解的速度快于遗传算法的速度。将粒子群算法和支持向量机结合起来,提出了一种粒子群支持向量机分类器,并通过在煤矿顶板状态分类检测中的应用,验证了此粒子群支持向量机分类器具有高效的分类能力。

In this paper, the penalty parameter C and kernel parameter σ of support vector machines is optimized using and genetic algorithm respectively. The simulation results show that the search speed of PSO is faster than genetic algorithms. And PSO has good global search, small calculated amount and simple operation. Then, the PSO support vector machine classifier is proposed. It is applied in the classification of coal mine roof failure. The results show that the PSO support vector machine classifier has efficient classification ability.