微波碳热还原低品位钛精矿过程中铁金属化率受到诸多因素的影响，其工艺参数难以寻优。针对微波碳热还原低品位钛精矿工艺过程中配碳量、还原温度、保持时间对铁金属化率的影响，采用响应面法和神经网络建立相应的响应面优化模型和一维卷积神经网络预测模型，对还原过程进行分析及工艺参数寻优。研究结果表明，铁金属化率随还原温度和保持时间的增加而增大，配碳量对还原铁金属化率的影响呈现先增加后降低的趋势。响应面法得出最佳工艺操作条件为还原温度1 091℃、保持时间76 min、配碳量10%，此条件下铁金属化率为97.672 5%。在工艺参数范围内，一维卷积神经网络模型能有效预测结果，为后续生产过程提供理论指导。

The conversion rate of iron in the process of microwave carbothermal reduction of low-grade ilmenite concentrate is affected by many factors, and the process parameters are difficult to be optimized.In view of the influences of the proportion of coke, reduction temperature and holding time on the metallization rate of iron in the process of microwave carbothermic reduction of low-grade ilmenite concentrate, a response surface optimization model and a one-dimensional convolution neural network predictive model are respectively built by means of response surface method and neural network, and used to analyze the reduction process and optimize the process parameters.Results show that the metallization rate of iron increases with the increase of reduction temperature and holding time.With the increasing proportion of coke, the metallization rate of iron increases first and decreases then.The optimum operating conditions obtained from response surface method are as follows:the reduction temperature is at 1, 091℃, the holding time is 76 min and the proportion of coke is 10%, under which the metallization rate of iron can reach 97.672 5%.Within the ranges of process parameters, using one dimensional convolutional neural network can effectively predict the results and provide theoretical guidance for the subsequent production process.