The generalised reduced gradient refinement was applied to optimise the constitutive constants obtained from hyperbolic-sine Arrhenius equation when describing the flow stress of two titanium alloys subjected to hot compression testing. The results showed that correlation coefficients improved from 0.96 and 0.98 to 0.99, while the average absolute relative error and the root mean square error reduced by more than 30%. The simple generalised reduced gradient refinement can be used to improve the prediction of flow stress when hyperbolic-sine Arrhenius equation or other phenomenological and physical models are used for describing hot working behaviour of metals and alloys.

当描述经受热压测试的两种钛合金的流应力时，使用广义的减小的梯度细化来优化从双曲线正弦Arrhenius方程获得的本构常数。结果表明，相关系数从0.96和0.98提高到0.99，而平均绝对相对误差和均方根误差降低了30％以上。当使用双曲线正弦Arrhenius方程或其他现象学和物理模型描述金属和合金的热加工行为时，可以使用简单的广义简化梯度细化来改进对流应力的预测。