提出了一种高效的ＢＰ神经网络学习算法，首次对神经网络输出结果进行自适应残差补偿，基于此提出炼铜转炉吹炼终点组合预报模型．经实践检验．本预报模型的预测结果具有较高的精度和较强的实时性，而且本模型具有较好的自学习能力，能有效指导生产实践。 An efficient BP neural network learning algorithm is proposed. For the first time, adaptive residual compensation is performed on the output of the neural network. Based on this, a combined prediction model for the end point of blowing converter is proposed. Practice test. The prediction results of this forecasting model have higher precision and stronger real-time performance. Moreover, this model has better self-learning ability and can effectively guide the production practice.