基于MIV-IPFA-ELM的矿井突水水源识别模型

Mine water inrush source identification model based on MIV-IPFA-ELM

为提高矿井突水水源识别准确率,提出一种基于MIV-IPFA-ELM的矿井突水水源识别模型。利用改进的探路者算法(IPFA)对极限学习机(ELM)的参数进行寻优,构建IPFA-ELM判别模型;采用平均影响值(MIV)方法计算各判别指标平均影响贡献率,根据计算结果剔除影响贡献率较低的判别指标;将筛选后的判别指标作为输入向量重新对模型进行训练,构建MIV-IPFA-ELM模型。以秦南矿井的实测水源数据为例,借助SMOTE算法生成类平衡实验数据集,并以该数据集验证构建模型的有效性,将实验结果与其他模型相比较。研究结果表明:MIV-IPFA-ELM模型的平均预测准确率为96.33%,远高于其他模型的预测准确率,均方误差为0.129,平均绝对误差为0.0625,较其他对比模型有明显的降低。

In order to improve the identification accuracy of mine water inrush source,a mine water inrush source identification model based on MIV-IPFA-ELM was proposed.Firstly,the improved Pathfinder algorithm(IPFA)was used to optimize the parameters of extreme learning machine(ELM),and the IPFA-ELM discriminant model was constructed.Secondly,the mean impact value method(MIV)was used to calculate the average impact contribution rate of each discriminant index,and the discriminant index with lower impact contribution rate was eliminated according to the calculation results.Finally,the discriminant indexes after screening were used as input vectors to re-train the model,and the MIV-IPFA-ELM model was constructed.Taking the measured water source data of Qinnan mine as an example,SMOTE algorithm was used to generate class balance experimental data set,and this data was used set to verify the validity of the model,and the experimental results were compared with other models.The results show that the average prediction accuracy of MIV-IPFA-ELM model is 96.33%,which is much higher than that of other models.The mean square error is 0.129 and the mean absolute error is 0.0625,which is significantly lower than that of other models.