改进的差分进化算法在冷连轧轧制规程优化中的应用研究

发布时间：2018-08-02 13:45

【摘要】：冷连轧机组的负荷分配是一个比较常见的基于非线性约束的多目标优化问题。轧制规程的设定关系到轧制产品的质量及生产能耗,制定合理的轧制规程可以在保证产品质量的同时,延长轧机使用寿命,减少故障发生概率,有效提高生产经济性。本文根据生产的实际需求,对传统的差分进化算法进行研究,提出一种全新的结合外部种群的自适应差分进化算法(Self-adaptive Differential Evolution Algorithm,SE-DE)对轧制规程这一多目标优化问题进行求解。首先,对传统的差分进化算法进行了分析,在原有的种群初始化部分引入一个外部种群进行反向学习,从而扩大了搜索空间,增加种群的多样性;其次,分析了各种自适应选择机制,针对传统算法中固定控制参数的局限性,设计了利用轮盘赌机制自适应的选择变异因子F的策略,提高算法对不同问题的适应性;最后,为了平衡算法的探索能力和开发能力,提出优化因子?的概念,通过对?值的判断来确定子代的交叉策略,两种交叉策略的分阶段应用,可以在提高SE-DE算法收敛性的同时有效避免搜索陷入局部最优的情况。为了验证所提算法在解决数学基准问题时的能力,选取包括传统DE算法在内的五种改进差分进化算法SADE,ODE,NDE和MDE-pBX,对CEC-2005的前14个测试函数分别进行30维、50维和100维的试验,仿真结果验证了SE-DE算法的优越性能。以某钢厂四辊五机架冷连轧生产线为例,对冷连轧生产线控制系统的组成、设备参数及工艺要求进行分析,并对轧制过程中相应的轧制参数数学模型进行详细的描述和分析。在此基础上,确立以预防打滑和等相对负荷为目标的多目标函数优化模型,将改进的差分进化算法SE-DE应用于冷连轧规程优化,并与传统差分进化算法的优化结果进行对比。通过对各机架压下量进行优化,仿真结果验证了所提算法能够对系统进行高效、快速的调节,对其他领域的多目标优化问题具有指导意义。
[Abstract]:Load distribution of cold tandem mill is a common multi-objective optimization problem based on nonlinear constraints. The setting of rolling schedule is related to the quality of rolling products and the energy consumption of production. Making a reasonable rolling schedule can extend the service life of rolling mill, reduce the probability of failure and improve the economy of production effectively while ensuring the quality of the rolling mill. In this paper, according to the actual demand of production, the traditional differential evolution algorithm is studied, and a new adaptive differential evolution algorithm (Self-adaptive Differential Evolution algorithm SE-DE) combined with external population is proposed to solve the multi-objective optimization problem of rolling schedule. Firstly, the traditional differential evolution algorithm is analyzed, and an external population is introduced to carry out reverse learning in the original population initialization part, which expands the search space and increases the diversity of the population. This paper analyzes various adaptive selection mechanisms, aiming at the limitation of fixed control parameters in the traditional algorithm, designs a strategy of adaptive selection of variation factor F using roulette mechanism to improve the adaptability of the algorithm to different problems. In order to balance the ability of exploration and development of the algorithm, an optimization factor is proposed. Through the concept of? Two kinds of crossover strategies can improve the convergence of SE-DE algorithm and avoid the situation of searching falling into local optimum. In order to verify the ability of the proposed algorithm in solving the mathematical benchmark problem, five improved differential evolution algorithms, SADEODENDE and MDE-pBX, including the traditional DE algorithm, are selected, and the first 14 test functions of CEC-2005 are tested on 30 D 50 and 100 D respectively. The simulation results verify the superior performance of SE-DE algorithm. Taking the four-roll five-stand cold continuous rolling line in a steel plant as an example, the composition of the control system, the equipment parameters and the technological requirements of the cold continuous rolling line are analyzed, and the corresponding mathematical model of the rolling parameters is described and analyzed in detail. On this basis, a multi-objective function optimization model with the aim of preventing skid and equal relative load is established. The improved differential evolutionary algorithm (SE-DE) is applied to the optimization of cold tandem rolling process, and the results are compared with those of the traditional differential evolutionary algorithm. The simulation results show that the proposed algorithm can adjust the system efficiently and quickly, and it is of great significance to the multi-objective optimization problem in other fields.
【学位授予单位】：燕山大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TG335.13;TP18

【相似文献】