金属力学性能质量预测及其控制  

Quality prediction model and control of the metallic mechanical properties

作　　者：王丹民[1] 李擎[2] 李华德[2] 

机构地区：[1]中冶京诚工程技术有限公司技术研究院,北京100176 [2]北京科技大学信息工程学院,北京100083

出　　处：《控制理论与应用》2007年第4期669-673,共5页Control Theory & Applications

摘　　要：为了在生产中能够对力学性能做出精确的预测,在用神经网建立起由工艺参数预测力学性能的质量模型后,又提出一种新的建模方法——逐层逼近法,从测试的结果看,后者预测精度明显高于前者.然后,对利用自适应逆控制方法实现对力学性能的在线控制进行了研究,仿真结果证明了该方法的有效性.In order to precisely predict the mechanical properties, the model, that could predict the mechanical properties of hot-rolled steel strip with the technological parameter, is established based on artificial neural network. A new method by applying the technology of layer-to-layer prediction and impending is also proposed. The test result shows that the precision of the latter prediction is higher than that of the former. Then, the real time control method of the mechanical properties is researched and simulated with adaptive inverse control. Simulation result shows that this control method is effective.

关 键 词：力学性能 人工神经网络 组织性能预测和控制 自适应逆控制 

分 类 号：TG115.5[金属学及工艺—物理冶金]