Abstract: In order to solve the problem of local optimization and low solving accuracy of the structural optimization of mulberry leaf picking machine, an improved real genetic algorithm is proposed. Firstly, a roulette selection operator based on order combination fitness function is given, the operator selects one of the two fitness functions through an adaptive parameter change on the basis of roulette, and then calculates the fitness value. Then, an improved heuristic crossover operator based on direction is designed, which not only preserves the influence of the best of the two parent individuals on the offspring individuals, but also increases the influence of the best of the population on the offspring individuals, so as to increase the possibility of crossover producing potential offspring. Then, the improved algorithm is applied to the optimization parameter design of rocker mulberry leaf picker, and the superiority of the algorithm is verified by simulation and comparison test with other algorithms, and the optimal parameter combination of the picker is obtained as follows, the speed of the walking structure is 24 mm/s, the angular speed of the flipping structure is 1.2 rad/s, and the speed of the picking structure is 440 mm/s. The performance of the whole machine is improved by 13% compared with that before optimization. Finally, field tests with the optimized parameter combination show that the performance of the mulberry leaf picker is improved by 10.9% and the error is less than 2.1%. It can be seen that the improved real genetic algorithm is an effective algorithm to optimize the parameters of the picker.

Key words: mulberry leaf picking machine, structure optimization, real coded genetic algorithm, roulette selection, heuristic crossover operator

摘要： 针对遗传算法在求解桑叶采摘机结构优化问题时容易陷入局部最优和求解精度低等问题，提出一种改进实数遗传算法。首先是给出一种基于序的组合适应度函数的轮盘赌选择算子，该算子在轮盘赌的基础上，通过一个自适应变化的参数在两种适应度函数中选择一个，再去计算适应度值；然后设计一种基于方向的改进启发式交叉算子，该算子既保留两个父代个体中较优个体对子代个体的影响，又增加种群中最优个体对子代个体的影响，提高交叉产生有潜力子代的可能性。接着将改进算法应用于摇杆式桑叶采摘机的优化参数设计中，通过与其他算法作仿真对比试验验证算法的优越性，获得采摘机最优参数组合:行走结构速度为24 mm/s、拨动结构角速度为1.2 rad/s、采摘结构速度为440 mm/s，并由运行结果可知整机性能与优化前相比提高13%。最后用优化得到的参数组合进行实地试验，结果显示桑叶采摘机性能提升10.9%，误差较小为2.1%。可见，所提改进实数遗传算法是优化采摘机参数的一种有效算法。

关键词: 桑叶采摘机, 结构优化, 实数遗传算法, 轮盘赌选择, 启发式交叉算子