一种改进算法在焦炉正常推焦过程中的应用

Application of an Improved Algorithm in the Normal Coke Pushing Process of Coke Oven

焦炉正常推焦过程中存在多个控制参数。通常人工经验法根据参数取值范围选取参数,该方法存在很大的随机性而且难以权衡多个性能指标使得正常推焦过程整体最优。为此,提出一种改进的多目标差分进化算法(IMODEA)。首先,采用了2个自适应性参数缩放比例因子和交叉概率完成变异和交叉操作。其次,提出一种改进的快速非支配排序策略(IFNSS)并利用IFNSS和拥挤度计算共同解决新一代种群选择问题。通过算法性能测试将IMODEA与NSGA-II作了对比,显示了IMODEA在Pareto最优解集空间分布性和收敛性方面更占优势。将IMODEA应用到焦炉正常推焦过程优化模型中,通过实验分析证明了算法的可行性。

There are several control parameters in normal coke pushing process of coke oven. Generally, artificial experience method selects parameters according to the scope of parameter values. However, this method not only has certain randomness in choosing parameters but also can＇t weigh every performance index to optimize the normal coke pushing process. Therefore, an improved multi-objective Differential Evolution algorithm（IMODEA） is proposed in this paper. First of all, two improved adaptive parameters, scaling factor and crossover probability, are employed to complete the mutation operation and crossover operation. Secondly, an improved fast non-dominated sorting strategy（IFNSS） is put forward to solve the problem of new generation population selection together with congestion degree calculation. Performance comparison of IMODEA to NSGA-II shows that IMODEA has an advantage in spatial distribution of Pareto optimal solution set and convergence of algorithm. IMODEA is applied to normal coke pushing process and its practicability is verified through experimental analysis.