论文部分内容阅读
由于模锻过程具有强时变性和非线性,因此精确控制模锻压机载荷预测至关重要。以7075铝合金模锻过程为例,提出了一种基于神经网络的模锻压机载荷在线建模方法。基于商业软件Deform-3D模拟了恒温恒速度工况下的载荷变化规律,根据获取的数据建立了初始神经网络模型。在实际模锻实验过程中,通过反向传播算法不断修正初始神经网络权值矩阵,以实现模型的在线更新。在50 t模锻实验台上进行实验,以验证所提方法的有效性。实验结果表明:所提出的在线建模方法可以准确预测复杂模锻工况下载荷的变化,与传统离线神经网络建模方法相比,其预测值更加准确,更能满足实际工程需求。
Due to the strong time-strain and non-linearity of die forging process, it is very important to accurately control the diecasting press load. Taking die forging process of 7075 aluminum alloy as an example, an on-line modeling method of die forging press load based on neural network was proposed. Based on the commercial software Deform-3D, the load variation under constant temperature and constant velocity was simulated. Based on the obtained data, an initial neural network model was established. In the actual die forging experiment, the initial neural network weight matrix is continuously corrected by the back propagation algorithm to realize the online update of the model. Experiments were conducted on a 50 t die forging test bench to verify the effectiveness of the proposed method. The experimental results show that the proposed online modeling method can accurately predict the change of load under complex forging conditions. Compared with the traditional offline neural network modeling method, the predicted value is more accurate and can meet the actual engineering requirements.