用改进遗传算法优化的带钢卷取温度预报模型

发布时间：2018-09-11 06:14

【摘要】：由于热轧带钢卷取温度控制过程存在强非线性,经典数学模型难以精确描述,我们采用遗传神经网络建立了卷取温度预报模型,并且通过改进的遗传算法优化了神经网络的权值.其中,提出了重新进化的思想,用"返祖"操作找回丢失的较优模式并将其耦合至下一代种群中,极大的提高了算法的收敛速度;分析了"种群解的空间跨度"和"基因段距离"对种群多样性的影响,用"优生"操作来推动算法从平面到多维空间的立体式搜索,以勘探和挖掘出更广、更优的寻优区间,并在种群进化后期,强力驱动算法收敛于全局最优.MFC(微软基类库)仿真结果表明:该卷取温度预报模型的收敛速度快、精度高,满足实时在线的控制要求,预报精度在±10?范围之内,能为卷取温度的前馈补偿控制提供可靠的参考数据,从而为进一步提高卷取温度的控制精度提供了新的途径.
[Abstract]:Because of the strong nonlinearity in the coiling temperature control process of hot strip, the classical mathematical model is difficult to describe accurately. We use genetic neural network to establish the coiling temperature prediction model, and optimize the weights of the neural network through the improved genetic algorithm. The convergence speed of the algorithm is greatly improved by coupling the optimal model to the next generation population, and the effects of "spans of population solutions" and "distances of gene segments" on the diversity of population are analyzed, and the "eugenic" operation is used to promote the three-dimensional search from plane to multi-dimensional space, so as to explore and mine a wider and better optimization region. The simulation results of MFC (Microsoft Base Class Library) show that the coiling temperature prediction model has the advantages of fast convergence speed, high accuracy, real-time and on-line control requirements, and the prediction accuracy is within the range of + 10?, which can provide reliable reference data for feedforward compensation control of coiling temperature. It provides a new way to further improve the control accuracy of coiling temperature.
【作者单位】：北京科技大学自动化学院;北华大学电气信息工程学院;北京科技大学钢铁流程先进控制教育部重点实验室;北华大学信息技术与传媒学院;
【基金】：北京市重点学科建设项目(XK100080537)资助~~
【分类号】：TG335.5;TP18

【参考文献】