MFOA-SVM在采煤工作面瓦斯涌出量预测中的应用  

Application of MFOA-SVM in Coalface Gas Emission Prediction

作　　者：王艳晖[1] 李国勇[1] 王炳萱 

机构地区：[1]太原理工大学信息工程学院,山西太原030024

出　　处：《矿业安全与环保》2016年第2期54-58,共5页Mining Safety & Environmental Protection

基　　金：国家自然科学基金项目(51075291)

摘　　要：针对新安煤矿采煤工作面瓦斯涌出量系统时变非线性特点,建立改进果蝇算法(MFOA)支持向量机(SVM)预测模型。利用FOA具有运算简单、收敛速度快、寻优精度高等优势来优化SVM核函数参数g、惩罚因子c和不敏感损失函数ε,但FOA也存在可能陷入局部最优的不稳定缺陷,则嵌入三维搜索、混沌优化、自适应变步长和最优保留策略进行改进,并利用Rosenbrock测试函数和采煤工作面瓦斯涌出量历史数据进行试验分析,结果表明:该模型预测平均相对误差为2.16%,比其他预测模型具有更高的预测精度、更快的收敛速度、更强的泛化能力,具有一定的实际应用价值。To counter to the time-varying non-linear characteristics of gas emission in Xin'an Coal Mine,a prediction model based on modified fruit fly optimization algorithm and support vector machine( MFOA-SVM) was established. The FOA( fruit fly optimization algorithm) has the advantages of simple operation,fast convergence speed and high optimization accuracy,which can be used to optimize the kernel function parameter g,the penalty factor c and insensitive loss function parameter ε of SVM,however,SVM has the unstable setback of trapping into local optimization,so it was modified by embedding the three-dimensional search,chaos optimization,adaptive variables step and the optimal retention strategy,and test analysis was carried out by using Rosenbrock function and the historical data of gas emission in coalfaces. The prediction results showed that the average relative error predicted with this model was 2. 16%,compared with other prediction models,it has the higher prediction accuracy,the faster convergence speed and the better application value.

关 键 词：瓦斯涌出量 MFOA-SVM 非线性 预测模型 

分 类 号：TD712[矿业工程—矿井通风与安全] TP18[自动化与计算机技术—控制理论与控制工程]