针对我国钢铁行业高耗能、高污染的问题,提出一种基于选择性神经网络集成算法的焦化工序能耗预测模型。通过对焦化工序实际能源产销数据进行训练得到神经网络学习器,再将不同抽样权重的样本数据通过神经网络模型,构成预测函数序列。最后,根据预测函数的预测效果进行剔除和保留,构成选择性集成预测模型。该方法能够将单一的弱学习算法通过集成提升为强学习算法,从而提高预测模型精度。实验结果表明,该方法在焦化工序投料配比与能耗预测的模型中取得了较好的效果,能够有效指导钢铁企业节能减排。 Aiming at the problem of high energy consumption and high pollution in China’s steel industry, this paper proposes a prediction model of energy consumption of coking process based on the selective neural network ensemble algorithm. The neural network learner is trained by actual energy production and sales data of the coking process, and the sample data of different sampling weights are passed through the neural network model to form the prediction function sequence. Finally, the predictive function of the predictive effect of the removal and retention, constitute a selective integrated forecasting model. This method can improve a single weak learning algorithm to a strong learning algorithm through integration, so as to improve the accuracy of the prediction model. The experimental results show that this method has achieved good results in the model of coking process feed ratio and energy consumption prediction, which can effectively guide the iron and steel enterprises to save energy and reduce emissions.