摘 要螺旋叶片是螺旋输送设备上的重要零部件，它的制造历史悠久。在生产螺

摘 要<br>螺旋叶片是螺旋输送设备上的重要零部件，它的制造历史悠久。在生产螺旋叶片的各种工艺中，目前最先进的是锥辊辗轧成形螺旋叶片。但是，这种工艺也有一个致命地弱点，那就是设备的工艺参数调整相当困难。目前采用这种螺旋叶片制造工艺的各国均采用“经验试轧法”来生产螺旋叶片，但是这种调整方法效率低下，浪费严重。因此，本文在螺旋叶片锥辊辗轧理论的基础上，提出以遗传算法（GA）来优化辗轧工艺参数，克服了“经验试轧法”的缺点。<br>本文在介绍了螺旋叶片的国内外研究现状和目前主要的螺旋叶片成形工艺，指出了螺旋叶片参数调整优化研究的重要意义基础上，首先介绍了螺旋叶片锥辊异面辗轧的理论，叙述了成形机理、变形区特点、咬入条件和成圆、成螺距的数学模型和一重要工艺参数喂入高度的确定方程。其次介绍了遗传算法的基本理论，确定了GA优化的运行参数和优化中的约束问题的处理，讨论了惩罚函数法在求解非线性方程中的应用。然后，把GA应用于辗轧工艺参数的优化，结合内点惩罚函数法处理约束，建立了优化的数学模型，目标函数和适应度函数，确定了具体操作过程，给出了优化流程，编制了优化程序。最后分别给出了标准GA优化和GA结合内点惩罚优化过程中各个工艺参数和目标函数的变化曲线，得到了最佳工艺参数，并通过实验，把轧出的叶片与标准叶片做比较，以检验GA对辗轧参数调整优化的效果，确定优化是否可行可靠。<br>研究结果表明：用标准GA优化工艺参数在120代左右时开始收敛，用其给出的最佳工艺参数试轧的螺旋叶片跟目标螺旋叶片在几何尺寸误差相对较大，还需要靠经验进一步微调才能达到要求，而用GA结合惩罚函数的优化在80～90代就已经开始收敛，用其给出的最佳工艺参数试轧出来的产品则完全符合产品的质量要求，这就克服了“经验试轧法”的缺点，达到了研究目的。同时通过误差分析我们发现GA优化后辗轧出的叶片的几何相对误差波动不大，这说明其具有一定的稳定性。因此用GA对辗轧参数的优化是切实可行可靠的，其对工厂的实际生产具有很大的指导作用。<br>关键词: 螺旋叶片 辗轧成形 遗传算法 参数优化

摘 要<br>螺旋叶片是螺旋输送设备上的重要零部件，它的制造历史悠久。 在生产螺旋叶片的各种工艺中，目前最先进的是锥辊辗轧成形螺旋叶片。 但是，这种工艺也有一个致命地弱点，那就是设备的工艺参数调整相当困难。 目前采用这种螺旋叶片制造工艺的各国均采用“经验试轧法”来生产螺旋叶片，但是这种调整方法效率低下，浪费严重。 因此，本文在螺旋叶片锥辊辗轧理论的基础上，提出以遗传算法（GA）来优化辗轧工艺参数，克服了“经验试轧法”的缺点。<br>本文在介绍了螺旋叶片的国内外研究现状和目前主要的螺旋叶片成形工艺，指出了螺旋叶片参数调整优化研究的重要意义基础上，首先介绍了螺旋叶片锥辊异面辗轧的理论，叙述了成形机理、变形区特点、咬入条件和成圆、成螺距的数学模型和一重要工艺参数喂入高度的确定方程。 其次介绍了遗传算法的基本理论，确定了GA优化的运行参数和优化中的约束问题的处理，讨论了惩罚函数法在求解非线性方程中的应用。 然后，把GA应用于辗轧工艺参数的优化，结合内点惩罚函数法处理约束，建立了优化的数学模型，目标函数和适应度函数，确定了具体操作过程，给出了优化流程，编制了优化程序。 最后分别给出了标准GA优化和GA结合内点惩罚优化过程中各个工艺参数和目标函数的变化曲线，得到了最佳工艺参数，并通过实验，把轧出的叶片与标准叶片做比较，以检验GA对辗轧参数调整优化的效果，确定优化是否可行可靠。<br>研究结果表明：用标准GA优化工艺参数在120代左右时开始收敛，用其给出的最佳工艺参数试轧的螺旋叶片跟目标螺旋叶片在几何尺寸误差相对较大，还需要靠经验进一步微调才能达到要求，而用GA结合惩罚函数的优化在80～90代就已经开始收敛，用其给出的最佳工艺参数试轧出来的产品则完全符合产品的质量要求，这就克服了“经验试轧法”的缺点， 达到了研究目的。 同时通过误差分析我们发现GA优化后辗轧出的叶片的几何相对误差波动不大，这说明其具有一定的稳定性。 因此用GA对辗轧参数的优化是切实可行可靠的，其对工厂的实际生产具有很大的指导作用。<br>关键词: 螺旋叶片 辗轧成形 遗传算法 参数优化

Abstract<br>Spiral blade is an important part of screw conveying equipment, and its manufacturing history is long. Among the various processes for producing spiral blades, the most advanced one is cone roll rolling. However, this process also has a fatal weakness, that is, it is very difficult to adjust the process parameters of the equipment. At present, the experience trial rolling method is adopted in the manufacturing process of spiral blade in many countries. However, this adjustment method is inefficient and waste seriously. Therefore, based on the theory of spiral blade cone roll rolling, genetic algorithm (GA) is proposed to optimize the rolling process parameters, which overcomes the shortcomings of "empirical trial rolling method".<br>This paper introduces the research status of spiral blade at home and abroad and the main forming technology of spiral blade at present, points out the important significance of parameter adjustment and Optimization Research of spiral blade, first introduces the theory of spiral blade conical roll non face rolling, describes the forming mechanism, the characteristics of deformation zone, the biting condition, the mathematical model of circle forming and pitch forming, and an important process parameter feeding The determination equation of height. Secondly, it introduces the basic theory of genetic algorithm, determines the operation parameters of GA optimization and the treatment of constraints in optimization, and discusses the application of penalty function method in solving nonlinear equations. Then, GA is applied to the optimization of rolling process parameters, and the mathematical model, objective function and fitness function of optimization are established, and the specific operation process is determined, the optimization process is given, and the optimization program is compiled. Finally, the change curves of process parameters and objective function in the process of standard GA optimization and GA combined with interior penalty optimization are given, and the optimal process parameters are obtained. Through experiments, the rolled blade is compared with the standard blade to test the effect of GA on rolling parameter adjustment and optimization, and determine whether the optimization is feasible and reliable.<br>The research results show that: the process parameters optimized by standard GA begin to converge at about 120 generations, and the geometric dimension error between the trial rolled spiral blade and the target spiral blade is relatively large, and it needs to be further adjusted by experience to meet the requirements. However, the optimization with GA combined with penalty function has already begun to converge in the 80-90 generation, and the optimal process given by this method has been used The products produced by parametric trial rolling fully meet the quality requirements of products, which overcomes the shortcomings of "empirical trial rolling method" and achieves the research purpose. At the same time, through error analysis, we find that the geometric relative error of the blade rolled by GA optimization has little fluctuation, which shows that it has a certain stability. Therefore, it is feasible and reliable to optimize the rolling parameters with GA, which has a great guiding role for the actual production of the factory.<br>Key words: spiral blade rolling forming genetic algorithm parameter optimization<br>