基于IGA-LSSVM的煤矿瓦斯涌出量预测模型研究

Study on Gas Emission Prediction Model Based on IGA-LSSVM

为有效提高煤矿瓦斯涌出量预测的准确性,进一步保障煤矿生产安全,提出经免疫遗传算法(IGA)优化的加权最小二乘支持向量机(LS-SVM),并用其建立煤矿瓦斯涌出量预测模型。首先针对瓦斯涌出量系统非线性、时变性、复杂性等特点,提出一种新的加权策略函数来改进LS-SVM。然后引入IGA,对改进的LS-SVM进行核参数δ和正则化参数γ寻优。最后,利用煤矿历史瓦斯涌出数据进行试验分析。结果表明,利用该模型预测的最大相对误差为2.763%,最小相对误差为0.705%,平均相对误差为1.329 8%,该模型较其他预测模型具有更快的收敛速度,更强的泛化能力和更高的预测精度。

In order to effectively improve accuracy of coal mine gas emission prediction, further ensuring the safety of coal mine, a gas emission prediction model based on IGA-LSSVM was built. In view of for gas emission system characteristics such as non-linearity, time depence, and complexity, a new strategy of weighting function was worked out to improve the LS-SVM. Then IGA was introduced to improve LS-SVM for nuclear and regularization parameters optimization. Finally the historical data on coal mine gas emission were used for test analysis. The prediction results show that by using the model to predict maximum rela- tive error is 2. 763% , the minimum relative error is 0. 705% and average relative error is 1. 329 8%. Compared with other predictive models, it has a faster convergence speed, better generalization ability and higher prediction precision.