基于改进万有引力算法的KELM瓦斯涌出量预测

Prediction of Gas Emission of KELM based on Improved Universal Gravitation Algorithm

针对瓦斯涌出量受其他因素的影响,并且存在着复杂的非线性关系,将核极端学习机与改进的万有引力算法相结合建立基于改进万有引力算法-KELM的瓦斯涌出量预测模型(IGSA-KELM瓦斯涌出量预测模型)。首先将输入样本作为KELM网络的输入量,然后采用改进的万有引力搜索算法对KELM网络的核参数和输出权值寻优,优化KELM网络的性能。测试结果表明,基于该方法预测的绝对瓦斯涌出量误差在0.1 m^3/min以内,提高了预测精度和预测效率。

In view of the influence of other factors on gas emission,and there is a complex non-linear relationship,a gas emission prediction model(igsa-kelm) based on the improved universal gravitation algorithm kelm is established by combining the nuclear extreme learning machine and the improved universal gravitation algorithm.Firstly,the input sample is taken as the input of kelm network,and then the core parameters and output weights of kelm network are optimized by the improved universal gravitation search algorithm to optimize the performance of kelm network.The test results show that the absolute gas emission error predicted based on this method is within 0.10 m^3/min,which improves the prediction accuracy and efficiency.