基于改进的自适应支持向量机建模的煤与瓦斯突出预测

Forecasting coal and gas outburst based on improved adaptive support vector machine

提出一种新的模糊隶属度函数对标准模糊支持向量机进行改进,然后运用自适应遗传算法对改进后的模糊支持向量机进行参数优选,得到一种新的AGAIFSVM模型,并且将提出的模型应用于煤与瓦斯突出预测。实验结果表明,所提出的模型比BP神经网络、标准支持向量机和模糊聚类有更高预测精度和更强的稳定性,具有较大的实用价值。

This paper proposed a new fuzzy function to improve on standard FSVM, also proposed a novel AGAIFSVM model. The model based on adaptive genetic algorithm to optimize the parameters of FSVM. In addition, applied the model to forecast coal and gas outburst. Experimental results show that AGAIFSVM model performs better than BP neural networks, standard SVM and fuzzy clustering method, implying that AGAIFSVM is very practical.