河蟹养殖全覆盖自动均匀投饲的轨迹规划与试验

河蟹养殖投饵喂料需要全塘均匀覆盖,目前主要靠人工驾驶或遥控船载投饲机在池塘中进行投喂,准确度和效率较低,难以保证投饲效果。针对以上情况,基于GPS（global positioning system）船载移动式自动投饲系统,研究了一种全覆盖轨迹规划方法,满足河蟹养殖池塘自动均匀投饲要求。该文利用轨迹规划系统建立池塘区域平面坐标系;以实际饵料累积密度与期望分布密度间的均方差最小作为衡量投饲均匀度的目标函数,求解出系统最优运行参数,生成往复遍历式自动投饲轨迹。仿真结果表明,在面积为2298.08 m2的不规则四边形池塘区域中,取饵料分布密度期望值为9 g/m2时,最优计算投饲系统运行参数对应的轨迹规划效果要优于传统经验估算参数。试验结果表明,当分别用两组参数进行自动投饲作业时,投饲船的实际航行轨迹与理论投饲轨迹都只有很小的偏移,与仿真结果具有一致性;同时,最优计算运行参数对应实际路径总长度、投饲路径长度、作业总时间、投饲时长、饵料分布密度均值、覆盖率等各项指标与仿真结果相比,相对误差分别为6.85%、2.71%、10.07%、10.41%、3.06%和1.87%,验证了全覆盖轨迹规划方法的可行性。该文可为河蟹养殖自动均匀投饲和其他水产养殖中需要沿池或全池自动均匀投饲轨迹规划研究提供重要参考。

Trajectory planning and test for all coverage, automatic and uniform feeding in river crab aquaculture

Abstract: Uniform feeding is needed for raising river crab on the whole pond. It mainly relies on artificial experience to determine the feeding amount and feeding route, and is always performed by workboat equipped with feeding device with manual driving or remote control on pond, which will result in strong random route, and low accuracy and efficiency and is difficult to guarantee the feeding effect. In view of the above situation, a full coverage trajectory planning method for automatic feeding system carried by workboat based on GPS (global positioning system) was proposed to meet the requirement of automatic and uniform feeding for raising river crab. In this paper, the quadrilateral pond area was selected on Google satellite map by using automatic feeding trajectory planning system based on GIS (geography information system), the GPS latitude and longitude coordinates of 4 vertices were converted to the plane coordinates with direct Gauss-Kruger projection calculation formula, and the working area was set by using safe distance and direction vector. A target function with minimum mean square error between expected distribution density and actual feed cumulative density was created to measure the feeding uniformity, and the optimal operation parameters were solved by using genetic algorithm. Feature point coordinates of trajectory planning were calculated by finding cross points between working area boundary and parallel lines along the longest side, the plane coordinates of feature points were converted to GPS latitude and longitude coordinates with inverse Gauss-Kruger projection calculation formula, and then reciprocating traverse trajectory for automatic feeding was generated and sent to automatic feeding system through the GPRS (general packet radio service) communication module. The simulation results showed that in an irregular convex quadrilateral area of 2 298.08 m2, when the expectation value of distribution density was 9 g/m2, the feeding performance indices of optimal calculation trajectory planning were better than that of traditional empirical estimation trajectory planning when feeding system worked with the parameters respectively. For the optimal calculation trajectory planning, corresponding total path length was 235.39 m, feeding path length was 195.16 m, total working time was 789.10 s, feed average density was 8.85 g/m2, and feeding coverage rate was 86.84%. Feeding contrast tests between traditional empirical estimation trajectory planning and optimal calculation trajectory planning were performed in Jiangsu University in April 2016. Results showed that experimental trajectories of workboat using the 2 groups of parameters both had only small offsets with the planned trajectories, which were consistent with the simulation results, and all feeding performance indices of optimal calculation trajectory planning were better than that of traditional empirical estimation trajectory planning. For the optimal calculation trajectory planning, corresponding practical total path length was 251.52 m, feeding path length was 200.45 m, total working time was 868.5 s, feeding time was 501.1 s, feed average density was 8.58 g/m2, and feeding coverage rate was 88.46%. And the relative error was 6.85%, 2.71%, 10.07%, 10.41%, 3.06% and 1.87% when comparing with the simulation results respectively, which showed the feasibility of the coverage trajectory planning method. Furthermore, trajectory planning verification simulation on another larger river crab pond of 15 057.56 m2 was performed, which corroborated that the automatic coverage trajectory planning system was also effective and applicable for large area river crab pond. The system could generate feeding trajectory directly and automatically when getting vertex coordinates of pond area, which would improve the accuracy and efficiency of trajectory planning, guarantee the uniform feeding and increase the economic benefits. This research provides technology reference for further study of trajectory planning for automatic uniform feeding on the whole pond in river crab aquaculture and other aquaculture for feeding along the pond or on the whole pond.