基于改进GWO-ELM的矿井突水水源识别算法

Recognition Algorithm of Mine Water Inrush Source Based on Improved GWO-ELM

针对现有矿井水源判别方法的局限性与煤矿开采水害事故高发性等问题,通过灰狼算法的优化方式对ELM极限学习机算法进行改进,建立改进GWO-ELM算法对顾北矿区的矿井样本水源进行识别,将归一化处理后的6项水化学离子指标作为网络输入向量,样本对应的水源层作为输出向量,通过GWO对网络初始权值阈值的更新迭代有效得到适用于矿井突水水源判别的单隐含层神经网络模型,优化后的网络模型的水源判别准确率高达92.3%,提高了网络结构的稳定性与鲁棒性,解决了矿井水源判别的低效率、低准确率突破等问题,对煤矿灾害防治工作中起到重大推进作用。

Aiming at limitations of existing discriminant method of mine water and coal mining accident high-risk sexual prob⁃lems such as water disasters,this paper uses the way of grey wolf optimizer to improve extreme learning machine algorithm,GWO-ELM algorithm for Gubei mining area of the mine water samples is established,which can identify the normalization process⁃ing after six measures of hydrochemistry ion as the network input vector,the water source layer corresponding to the sample is used as the output vector,initial weights of network through GWO comes update threshold iteration effectively get water suitable for mine water inrush criterion of single hidden layer neural network model,the optimized network model of water discriminant accuracy is 92.3%.It improves the stability and robustness of the network structure,solves the problems of low efficiency and low accuracy breakthrough in the discrimination of mine water source,and plays an important role in promoting the prevention and control of coal mine disasters.