基于神经网络的轧制力预报模型

采用了BP神经网络对热轧无缝钢管穿孔过程中轧制力进行预测，在BP算法学习过程中引入了附加动量法和自适应学习速率，结合Levenberg-Marquardt优化方法，加快了学习时的收敛速度，试验证明，取得了良好的学习和测试效果。研究结果可为斜轧穿孔工具设计和工艺调整提供一定的理论依据，对解决生产中出现的实际问题、开发新产品和新工艺提供有效的指导。

Rolling Force Forecast Model Based on BP-Neural Network

BP neural network is adopted to forecast rolling force during producing hot-rolled seamless steel tubes, BP arithmetic of introduction of additional momentum and adaptive learning rate, combining Levenberg-Marquardt optimization method importing inertia gene, accelerates the speed of learning and gets the nicer result. Results of this study not only can be used on the design of tools, but also is significative for equipment adjustment and process parameters optimization. What is more, it can be as a theories guide in solving current piercing problems, exploiting new seamless steel tube and designing new process, simultaneity, it is also have some use for reference function to solve the similarity problems in seamless steel tube production.