| 引用本文: | 孙铁军,杨卫东,程艳明,段凤云,弥洪涛.用改进遗传算法优化的带钢卷取温度预报模型[J].控制理论与应用,2015,32(8):1106~1113.[点击复制] |
| SUN Tie-jun,YANG Wei-dong,CHENG Yan-ming,DUAN Feng-yun,MI Hong-tao.Improved genetic algorithm for optimizing prediction model of strip coiling temperature[J].Control Theory and Technology,2015,32(8):1106~1113.[点击复制] |
|
| 用改进遗传算法优化的带钢卷取温度预报模型 |
| Improved genetic algorithm for optimizing prediction model of strip coiling temperature |
| 摘要点击 1970 全文点击 2193 投稿时间：2015-01-14 修订日期：2015-08-27 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/CTA.2015.50041 |
| 2015,32(8):1106-1113 |
| 中文关键词 卷取温度 遗传算法 返祖 基因段距离 优生 |
| 英文关键词 coiling temperature genetic algorithm atavism gene segment distance eugenics |
| 基金项目 北京市重点学科建设项目(XK100080537)资助. |
|
| 中文摘要 |
| 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. |
| 英文摘要 |
| Because of the high nonlinear features in the process of the hot rolled strip coiling temperature, it is difficult to use classical methods to build up an accurate mathematical model; we in this paper build for the strip coiling temperature a prediction model based on genetic neural network, and optimize the weights of the neural network through the improved genetic algorithm. In this scheme, we introduce the idea of re-evolution and employ the ‘atavism’ operation to retrieve the superior operation mode that has been lost and couple it into the next generation of population, to increase the convergence speed of the algorithm. We also analyze the impact on population diversity from the ‘space spans of population solution’ and the ‘gene segment distance’. Moreover, we use ‘Eugenics’ operation to extend the algorithm search from a plan to a solid space to explore and excavate a broader, superior optimization interval. In the later section of the evolution process, the algorithm is highly driven to converge to the global optimum. Simulation results of MFC (Microsoft Foundation Classes) show that this prediction model of strip coiling temperature is with the advantage of fast convergence and high precision, satisfying the requirements of the real-time online control with a prediction accuracy within the range of ±10℃. Thus, it can provide with reliable reference data in the feedforward compensation control for coiling temperature, and offers a new way to further enhance the control precision of coiling temperature. |
|
|
|
|
|