基于VMM和DNiGA的直升机多林区植保作业航线规划  

Route planning for multi-forest areas of helicopter plant protection operations based on vector modeling method and double-layer nested improved genetic algorithm

作　　者：方树平 茹煜[1] 胡晨明 刘洋洋 刘彬 李建平 夏达明 Fang Shuping;Ru Yu;Hu Chenming;Liu Yangyang;Liu Bin;Li Jianping;Xia Daming(College of Mechanical and Electronic Engineering,Nanjing Forestry University,Nanjing 210037,China;College of Mechanical Engineering,Anhui Science and Technology University,Fengyang 233100,China;School of Engineering,Anhui Agricultural University,Hefei 230036,China)

机构地区：[1]南京林业大学机械电子工程学院,南京210037 [2]安徽科技学院机械工程学院,凤阳233100 [3]安徽农业大学工学院,合肥230036

出　　处：《农业工程学报》2022年第1期70-79,共10页Transactions of the Chinese Society of Agricultural Engineering

基　　金：国家重点研发项目(2018YFD0600202-04);安徽省重点研发计划项目(202104a06020001)。

摘　　要：直升机在现代林业病虫害防治中扮演重要角色,合理的航线规划对直升机航空施药至关重要。该研究针对直升机小坡度多片作业林区的情况,开发了一种适用于单个林区施药航线规划的向量建模法(VectorModeling Method,VMM)和适用于多林区间调度航线规划的双层嵌套遗传算法(Double Nested improved Genetic Algorithm,DNiGA)。VMM法以减少单林区间掉头次数和多余覆盖率为规划航线目标,DNiGA算法以外层遗传算法求解调度航线,内层遗传算法作为协助求解外层遗传算法适应度的算子,求解考虑各林区进出入点的区域间最优调度路径。在自建多林区环境下进行算法验证。对VMM和Slice算法进行对比,在3种喷幅下,VMM法的掉头次数均明显少于Slice算法,多余覆盖率分别为2.31%、5.28%和6.64%。对DNiGA与分步式IGA-GA算法进行比较,喷幅10、20、30 m时DNiGA算法求得区域间最优调度航线长度分别为5 006.09、4 644.97和5 295.47 m,较IGA-GA分别缩短了11.24%、5.11%和6.59%。R44直升机多林区植保作业试验表明,VMM和DNiGA算法规划的航线比人工经验法的掉头次数减少4次,多余覆盖率降低27.8%,施药航线长度缩短6.15%,区域间调度航线长度缩短16.41%。该算法适用于直升机多林区植保作业航线规划,可减少掉头次数,降低多余覆盖率,缩短区域间调度航线长度,对实现智慧农林植保作业具有一定价值。Forest pests and diseases have frequently caused huge economic losses in China. Helicopter spraying of pesticides has been one of the most important ways to pest control in modern forestry. Reasonable route planning is a high demand for helicopters during spraying operations in multiple forest areas. However, the current route planning still fully relies on manual experience, resulting in too many turnarounds, and large extra coverage areas. It is very necessary to optimize route planning for higher operational efficiency in an environmentally friendly way. In this study, an optimal helicopter route planning was developed for spraying pesticide operations in multiple forest areas. The routes were divided into two parts: intraregional application routes for individual forest areas and interregional dispatch routes between multiple forest areas. In the application routes, a Vector Modeling Method(VMM) was developed to reduce the number of turnarounds and extra coverage rate. A multiple objective genetic algorithm NSGA-II was used to find the Pareto optimal set, and then a manual decision was made on the application route plan. A Double-layer Nested improved Genetic Algorithm(DNiGA) was developed for the inter-regional dispatch route planning of multiple forests, in which the outer layer GA was used to solve the inter-regional scheduling routes, and the inner layer GA served as an operator to solve the adaptation of the outer layer GA. As such, the inter-regional optimal scheduling routes were derived suitable for the multiple forest areas. Taking Thunderbird R44 aircraft as an example, a multiple forest area environment was created to verify the model. Three spray widths of 10, 20, and 30 m were selected to plan the operating routes. The VMM and the popular Slice were used to plan the application routes. It was found that the number of turnarounds was significantly reduced less than that of Slice for all three spray widths. The extra coverage rate of the VMM was only 2.31% and 6.64% at the spraying width of 10 and

关 键 词：直升机 算法 植保 航线规划 施药航线 调度航线 

分 类 号：S2523[农业科学—农业机械化工程]