(哈尔滨工业大学 材料科学与工程学院，哈尔滨 150001)
摘 要: TA15钛合金经过不同条件的热约束变形之后进行金相观察，获得了工艺(温度、应变、应变速率、冷却方式)和组织(初生α相含量、初生α相尺寸、初生α相长径比)参数数据，分别以这些数据作为输入和输出，建立了结构为4×6×8×3的BP人工神经网络。研究结果表明：所建立的网络可以很好地反映出材料的工艺−组织之间的关系并且具有一定的精度，网络模型可以用来预测不同变形条件下TA15钛合金的组织，且对于TA15钛合金的实际生产具有有效的指导作用。
关键字: TA15钛合金；初生α相；BP人工神经网络；三维曲面
（School of Materials Science and Engineering, Harbin Institute of Technology, Harbin 150001, China）
Abstract: After hot restriction experiments, processing parameters (temperature, strain, strain rate, cooling) and microstructures(firstborn α quantity, size, aspect ratio)of TA15 titanium alloy were obtained through microscopic examination, a 4×6×8×3 BP artificial neural network was built with these parameters as neurons of its input and output layer. The results show that the model can reflect the relationships finely and has certain accuracy. The model can be used to predict the microstructures of TA15 titanium alloy. Meanwhile, it can also serve as a guide for the heat treatment of TA15 titanium alloy.
Key words: TA15 titanium alloy; firstborn α phase; BP artificial neural network; three dimensional shaded surface
