摘要
利用MATLAB软件建立了反映材料热变形本构关系的神经网络模型,该模型中采用遗传算法优化其权值和阈值提高了网络收敛的稳定性。并采用Themecmastor-Z型热加工模拟试验机上进行的TC11钛合金等温恒应变速率压缩试验获得的试验数据进行训练,建立了TC11钛合金热变形本构关系的BP神经网络模型,并进行了预测,预测误差小于10%。
A ANN model was established by use of the MATLAB software to reflect the constitutive relation in the hot deformation process, and the weights and bias were optimized by genetic algorithm to improve the stability of convergence in this paper. And the BP model of TC11 alloy constitutive relation in hot deformation process was established by use of the test data of performing isothermal constant-strain-rate compression tests on the Themecmastor-Z Thermal Simulator. Predicting of the model was performed , the results show predicting error is less than 10%.
出处
《山西冶金》
CAS
2008年第3期13-15,共3页
Shanxi Metallurgy
基金
山西省自然科学基金资助项目(2008011045)