基于GA-ELM浆体管道输送临界流速预测模型研究

针对浆体管道输送临界流速预测难度大、精确度低等技术难题，提出了基于极限学习机(ELM)的临界流速预测模型，用训练集对模型进行训练，以验证集预测值的均方误差作为适应度函数，利用遗传算法(GA)对ELM模型参数进行优化，应用优化得到的ELM模型对预测集进行预测。以某矿山为例，模型参数优化结果如下:隐含层节点数L为400，输入权值ai、偏置向量bi最优组合下预测结果适应度为0.0201。采用优化的ELM模型对预测集进行预测，预测结果的最大相对误差x=3.96%，平均相对误差y=1.58%，对比BP神经网络(x=12.95%)和SVM模型(x=3.19%)，表明ELM模型更加精确、高效。

Study on prediction model of critical flow velocity in slurry pipeline transportation based on GA-ELM

Considering the technical problems of great difficulties and low accuracies in predicting the critical flow velocity of slurry pipeline transportation, a prediction model of critical flow velocity based on extreme learning machine (ELM) was proposed. The training set was used to train the model, and the mean square error of the validation set value was selected as the fitness function. Then the genetic algorithm (GA) was used to optimize the parameters of ELM model. The optimized ELM model was used to predict the forecast set. Taking a certain mine as example, the optimized parameters of the model were as follows: the hidden layer nodes L was 400, and under the optimal combination of the input weights ai and the offset vectors bi, the fitness of prediction results was 0.0201. The maximum relative error x= 3.96%, with an average relative error y= 1.58%. Compared with BP neural network(x=12.95%) and SVM model(x=3.19%), the ELM model was more accurate and efficient.