变量喷药自适应神经模糊控制器设计与仿真

为了减少除草剂用量,采用变量喷施除草剂方式进行除草.根据分别建立的杂草面积、喷药机械行驶速度与喷药量关系模型,得知杂草面积和喷药机械行驶速度是影响变量喷施效果的主要因素.为了获取喷药量与车速及杂草面积关系试验数据,设计了室内变量喷药试验台,使用DSP处理器及编码器分别得到杂草面积及喷药机械前进速度信息.结合所获试验数据,设计了一种基于自适应神经模糊推理（ANFIS）的双输入、单输出控制器.对控制器设计过程中输入输出变量的选取、隶属函数的选择及控制器的训练等进行了研究,数据经过30次训练后误差为1.47×10^-5.对控制器的速度采集、串行通信、电磁阀驱动等硬件电路及模糊控制软件流程,进行了设计.在Matlab中建立了自适应神经模糊控制仿真模型,仿真结果表明：在喷头打开时间为0.2 s,喷药机械速度为0～1 m/s,杂草面积在0～100 cm^2时,控制器可自动调节喷药量在0～4 mL变化.与采用传统模糊控制方式相比,该控制器自适应性强,具有较好的应用前景.

Design and simulation of variable weed spraying controller based on adaptive neural fuzzy inference system

To reduce the amount of herbicide,variable herbicide spraying measure was used to control weeds.According to the respectively established relation model between weed size,spraying machine speed and spraying volume,it is found that the main factors that affect variable spraying were weed size and spraying machine speed.In order to obtain the experimental data about the relationship among spraying volume,spraying machine speed and weed size,an indoor variable spraying test rig was designed in which DSP processors were used to obtain weeds size information and encoder was used to obtain mechanical speed information.Combined with the experimental data obtained in the test stand,a double input,single output controller based on adaptive neural fuzzy inference system(ANFIS) was designed.The selection of input and output variables,the membership functions choice and the controller training were studied.The data error was 1.47×10-5 after 30 times training.Speed acquisitions,serial communication,the electromagnetic valve driving hardware structure and fuzzy control software process of the controller were designed.The ANFIS simulation model was established in Matlab.The simulation results show that when the opening time of nozzle is 0.2 s,the spraying machine speed is between 0-1 m/s,the weed size is between 0-100 cm2.The controller can change the spraying amount from 0-4 ml automatically.Compared to the traditional fuzzy control method,the controller is self-adaptable and has a good application prospect.