The simulation of a control system for the longitudinal axis of the rotary or fixed-wing unmanned aerial vehicles(UAVs)is demonstrated in this study.The control unit includes design considerations of two controllers t...The simulation of a control system for the longitudinal axis of the rotary or fixed-wing unmanned aerial vehicles(UAVs)is demonstrated in this study.The control unit includes design considerations of two controllers to provide robust stability,tracking of the proposed linear dynamics,an adequate set of proportional-integral-derivative(PID)controller gains,and a minimal cost function.The PID control and linear quadratic regulator(LQR)with or without full-state-observer were evaluated.An optimal control system is assumed to provide fast rise and settling time,minimize overshoot,and eliminate the steady-state error.The effectiveness of this approach was verified by a linear model of the UAV aircraft in the semi-dynamic simulation platform of Matlab/Simulink,in which the open-loop system was assessed in terms of flight robustness and reference tracking.The experimental results show that the proposed controllers effectively improve the configuration of the control system of the plant,maintain the sustainability of the dynamic flight model stability,and diminish the flight controller errors.The LQR provides robust stability,but it is not optimal in the transient phase of particular plant output.The PID control system can adjust the controller’s gains for optimal hovering(or stable slow flight)and is especially useful for the tracking system.Finally,comparing aircraft stability using PID and LQR controllers shows that the latter has less overshoot and a shorter settling time;in addition,all proposed controllers can be practically deployed as one UAV’s system,which can be handled as an exemplary model of the UAV flight management system.展开更多
In order to solve the problems of insufficient training equipment,relatively lack of curriculum resources and single teaching means in the teaching of UAV(unmanned aerial vehicle)applied technology major,this paper st...In order to solve the problems of insufficient training equipment,relatively lack of curriculum resources and single teaching means in the teaching of UAV(unmanned aerial vehicle)applied technology major,this paper studies the application of MR(Mixed Reality)in UAV applied technology major teaching,with the teaching of UAV agriculture&forestry plant protection curriculum as the carrier.The study will solve the pain points in teaching,improve the teaching ability and teaching information level,and increase the talent training quality of UAV,agriculture&forestry plant protection and related majors.Furthermore,it will create a protective,interactive,remote and scalable teaching experience for stu-dents,which can improve the teaching effect and reduce the teaching cost.展开更多
杂草在作物生长初期受环境变化影响快速扩散,严重压缩作物生长环境。为有效管理农田并准确获取杂草群落扩散位置和生长状况,采用深度学习技术,基于卷积长短期记忆网络(Convolutional long short term memory,ConvLSTM)模型及多特征融合...杂草在作物生长初期受环境变化影响快速扩散,严重压缩作物生长环境。为有效管理农田并准确获取杂草群落扩散位置和生长状况,采用深度学习技术,基于卷积长短期记忆网络(Convolutional long short term memory,ConvLSTM)模型及多特征融合对农田杂草群落扩散准确预测,通过无人机获取具有时间序列的数字正射图像(Digital orthophoto map,DOM)数据,数据预处理后,优化土壤调节植被指数(Optimizing soil adjustment vegetation index,OSAVI)阈值法,构建多种输入特征制作数据集。将ConvLSTM模型与多种输入特征融合并对模型进行堆叠优化,构建多特征融合卷积长短期记忆网络(Multi-feature convolutional long short term memory networks,MF-ConvLSTM)模型,实现多步预测,使用制作数据集进行网络训练,综合对比MF-ConvLSTM、ConvLSTM、深度神经网络(Deep neural network,DNN)、全连接长短期记忆网络(Fully-connected long short term memory networks,FC-LSTM)4个模型。结果表明,构建的MF-ConvLSTM模型预测效果较好,其综合性能优于ConvLSTM、DNN和FC-LSTM,均方误差(Mean square error,MSE)值为0.0191,较传统FC-LSTM模型下降0.0087、POD提高0.0702、CSI提高0.0583、FAR降低0.0727。在不同覆盖度和降雨量条件下,MF-ConvLSTM模型杂草群落扩散预测结果较为平均,拥有较稳定MSE值及预测精度,体现模型较好的鲁棒性。此外,根据试验可知特征输入和预测步长对MF-ConvLSTM模型有不同程度影响。研究提出MF-ConvLSTM模型能自适应学习短期时空依赖关系,在多特征共同输入和短期预测步长情况下达到最佳性能。研究为准确获取农田杂草群落扩散位置和生长状况提供思路和方法,也可为后续农田精准除草和制作杂草处方图提供参考。展开更多
文摘The simulation of a control system for the longitudinal axis of the rotary or fixed-wing unmanned aerial vehicles(UAVs)is demonstrated in this study.The control unit includes design considerations of two controllers to provide robust stability,tracking of the proposed linear dynamics,an adequate set of proportional-integral-derivative(PID)controller gains,and a minimal cost function.The PID control and linear quadratic regulator(LQR)with or without full-state-observer were evaluated.An optimal control system is assumed to provide fast rise and settling time,minimize overshoot,and eliminate the steady-state error.The effectiveness of this approach was verified by a linear model of the UAV aircraft in the semi-dynamic simulation platform of Matlab/Simulink,in which the open-loop system was assessed in terms of flight robustness and reference tracking.The experimental results show that the proposed controllers effectively improve the configuration of the control system of the plant,maintain the sustainability of the dynamic flight model stability,and diminish the flight controller errors.The LQR provides robust stability,but it is not optimal in the transient phase of particular plant output.The PID control system can adjust the controller’s gains for optimal hovering(or stable slow flight)and is especially useful for the tracking system.Finally,comparing aircraft stability using PID and LQR controllers shows that the latter has less overshoot and a shorter settling time;in addition,all proposed controllers can be practically deployed as one UAV’s system,which can be handled as an exemplary model of the UAV flight management system.
基金Supported by Vocational Education Reform and Innovation Project of Ministry of Education(HBKC217166,HBKC217168)Teaching Reform Project of Agricultural Specialty Teaching Steering Committee of Higher Vocational Education in Guangdong Province(YNYJZW2019YB09)+1 种基金Special Higher Vocational Enrollment Expansion Project of Teaching Reform Research and Practice Pro-ject in Guangdong Province(JGGZKZ2020141)Special Fund for Rural Revitalization Strategy of Huizhou in 2021(2021SC010502002)
文摘In order to solve the problems of insufficient training equipment,relatively lack of curriculum resources and single teaching means in the teaching of UAV(unmanned aerial vehicle)applied technology major,this paper studies the application of MR(Mixed Reality)in UAV applied technology major teaching,with the teaching of UAV agriculture&forestry plant protection curriculum as the carrier.The study will solve the pain points in teaching,improve the teaching ability and teaching information level,and increase the talent training quality of UAV,agriculture&forestry plant protection and related majors.Furthermore,it will create a protective,interactive,remote and scalable teaching experience for stu-dents,which can improve the teaching effect and reduce the teaching cost.
文摘杂草在作物生长初期受环境变化影响快速扩散,严重压缩作物生长环境。为有效管理农田并准确获取杂草群落扩散位置和生长状况,采用深度学习技术,基于卷积长短期记忆网络(Convolutional long short term memory,ConvLSTM)模型及多特征融合对农田杂草群落扩散准确预测,通过无人机获取具有时间序列的数字正射图像(Digital orthophoto map,DOM)数据,数据预处理后,优化土壤调节植被指数(Optimizing soil adjustment vegetation index,OSAVI)阈值法,构建多种输入特征制作数据集。将ConvLSTM模型与多种输入特征融合并对模型进行堆叠优化,构建多特征融合卷积长短期记忆网络(Multi-feature convolutional long short term memory networks,MF-ConvLSTM)模型,实现多步预测,使用制作数据集进行网络训练,综合对比MF-ConvLSTM、ConvLSTM、深度神经网络(Deep neural network,DNN)、全连接长短期记忆网络(Fully-connected long short term memory networks,FC-LSTM)4个模型。结果表明,构建的MF-ConvLSTM模型预测效果较好,其综合性能优于ConvLSTM、DNN和FC-LSTM,均方误差(Mean square error,MSE)值为0.0191,较传统FC-LSTM模型下降0.0087、POD提高0.0702、CSI提高0.0583、FAR降低0.0727。在不同覆盖度和降雨量条件下,MF-ConvLSTM模型杂草群落扩散预测结果较为平均,拥有较稳定MSE值及预测精度,体现模型较好的鲁棒性。此外,根据试验可知特征输入和预测步长对MF-ConvLSTM模型有不同程度影响。研究提出MF-ConvLSTM模型能自适应学习短期时空依赖关系,在多特征共同输入和短期预测步长情况下达到最佳性能。研究为准确获取农田杂草群落扩散位置和生长状况提供思路和方法,也可为后续农田精准除草和制作杂草处方图提供参考。