基于ANN的TC18钛合金β相区流动应力修正及加工性能研究

Flow stress correction and study on processing performance of TC18 titanium alloy inβphase region based on ANN

热压缩实验中的摩擦效应和变形温升会导致测得的流动应力不准确,常规流动应力修正时将温升引起的应力差值简化为温度的线性函数,存在较大的误差。为此,建立了考虑摩擦效应和变形温升效应的人工神经网络(ANN)模型。采用Gleeble-3500热模拟试验机获得了TC18钛合金在β相区的流动应力曲线,基于摩擦修正方法与变形温升计算方法,对流动应力、变形温度进行了数据校正并将其作为ANN模型的训练集,得到集成摩擦-温升修正的ANN模型。对比实验结果与ANN模型预测结果,发现摩擦效应和变形温升对流动应力影响显著,ANN模型预测的流动应力曲线与微观组织更加吻合,表明了ANN模型的准确性。进一步,基于ANN模型的预测结果建立了功率耗散图,并结合微观组织表征分析了合金热变形时的组织演变与加工性能。功率耗散因子η受变形温度和应变速率的影响显著,在950~990℃、0.01~0.001 s^(-1)条件下η值最高,此时合金发生了充分的动态回复和再结晶,热加工性能最好。研究结果为流动应力的修正方法提供了参考,对TC18钛合金β相区的热加工工艺参数制定具有重要的指导意义。

The friction effect and temperature rise of deformation in hot compression experiment could lead to the inaccuracy of the measured flow stress,and when the conventional flow stress is corrected,the stress difference caused by temperature rise is simplified as a linear function of temperature,which has a large error.Therefore,an artificial neural network(ANN)model considering the friction effect and temperature rise of deformation was established,and the flow stress curve of TC18 titanium alloy in theβ-phase region was obtained by thermal simulation experiment machine Gleeble-3500.Based on the friction correction method and the calculation method of temperature rise due to deformation,the data of flow stress and deformation temperature were corrected and used as the training set of the ANN model,and the ANN model with integrated friction-temperature rise correction was obtained.The comparison between the experimental and predicted results by the ANN model reveals that the friction effect and temperature rise of deformation have a significant influence on the flow stress,and the flow stress curves predicted by the ANN model are more consistent with the microstructure,which indicates the accuracy of the ANN model.Furthermore,a power dissipation diagram was established based on the ANN model prediction results,and the microstructure evolution and processing performance of the alloy during hot deformation wrere analyzed in combination with microstructure characterization.The power dissipation factorηis significantly affected by the deformation temperature and strain rate,the highest value ofηis found at 950-990℃and 0.01-0.001 s^(-1),at which the dynamic recovery and recrystallization of the alloy take place,and the hot processing performance is the best.Thus,the research results provide a reference for the correction method of flow stress,and are of great significance for the formulation of hot processing process parameters for TC18 titanium alloy inβ-phase region.