力学性能预测模型及其控制的研究  

STUDY ON QUALITY PREDICTION MODEL AND CONTROL OF THE MECHANICAL PROPERTIES

作　　者：王丹民[1] 李华德[1] 李擎[1] 

机构地区：[1]北京科技大学信息工程学院,北京100083

出　　处：《机械强度》2006年第6期878-882,共5页Journal of Mechanical Strength

摘　　要：首先介绍神经网络建模特点,利用神经网络建立起由工艺参数预测力学性能的质量模型,及根据力学性能要求优化工艺参数的逆质量控制模型,预测效果图显示该模型的预测精度较高。然后,利用质量预测模型分析卷取温度对屈服强度的影响,并利用自校正PID(proportionintegral differentiation)控制实现力学性能的控制,仿真结果证明该方法的有效性。The characteristics of establishing a model based on artificial neural network were introduced firstly. The quality model, that could predict the mechanical properties of hot-tiled steel strip with the technological parameter, and the reverse quality control model, that could optimize the tecbnological parameter with the mechanical properties, were established by applying the technology of artificial neural network. With the quality prediction model, stress yield effected by coiling temperature was analyzed. The control of mechanical properties based on self-adaptive PID（proportion integral differentiation）was realized and simulated finally.

关 键 词：力学性能 热轧带钢 神经网络 自校正PID(proportion INTEGRAL differentiation)控制 组织性能预测和控制 

分 类 号：TG335.5[金属学及工艺—金属压力加工] TP183[自动化与计算机技术—控制理论与控制工程]