神经元网络应用于圆坯连铸结晶器传热计算

Application of neural networks to heat transfer calculation in a round billet continuous casting mould

摘要: 以圆坯连铸结晶器温度实测数据为基础,研究应用神经元网络求解结晶器传热反问题的可行性,并尝试将神经元网络与数值计算结合用于结晶器传热计算,探索针对实测数据的结晶器传热在线计算方法.结果表明,数值计算结果与实测温度符合较好,可真实反映实际生产中圆坯结晶器传热的非均匀特性.

Abstract: Based on the measured data of mould temperature during continuous casting of round billets, the feasibility of using neural networks to solve the inverse problem of heat transfer of a continuous casting mould was discussed. The combination of neural networks with numerical simulation was proposed to calculate the heat transfer of the mould. The results show that the temperature and heat flux field calculated are in agreement with the experimental ones, which could correctly reveal the characteristics of non-uniform heat transfer around the mould circumference.