应用人工神经网络建立TC11钛合金化学元素与力学性能关系模型

Modeling of Chemical Elements and Mechanical Property for TC11 Titanium Alloy Based on the Artificial Neural Network

摘要:

在TC11钛合金大量实验数据的基础上，应用人工神经网络建立TC11钛合金的化学元素与力学性能关系模型。模型的输入参数包括Al、Mo、Zr、Si、Fe、C、O、N和H共9种化学元素；输出为常规力学性能指标 (抗拉强度、屈服强度、延伸率和断面收缩率)。运用未知数据样本对已建立神经网络模型的预测能力进行检验，并以Al、Mo、Zr和C元素为研究对象，利用该模型分析TC11钛合金化学元素对力学性能的影响规律。结果表明：网络的预测值与实验值的相对误差均在10%以内，说明所建立的神经网络预测模型具有较精确的预测能力，而且能够清楚地反映出该合金化学元素与力学性能之间的非线性关系

Abstract:

Based on a large amount of experimental data, the relationship model of chemical elements and mechanical property for TC11 titanium alloy has been developed using artificial neural network. The input parameters of this model were 9 kinds of elements, including Al, Mo, Zr, Si, Fe, C, O, N and H. The mechanical properties were used as output parameters, including ultimate tensile strength, yield strength, elongation and reduction of area. The prediction capability of the established model was tested by the unseen data sample. Additionally, the effect of chemical elements (Al、Mo、Zr and C) on the mechanical property was studied using the present model. It is found that the relative errors between predicted and experimental values all within 10%, indicating that the neural network model possesses excellent prediction capability. With the help of the trained ANN model, the nonlinear relationship of chemical elements and mechanical property can also be clearly presented.