融合视觉多信息的网络化控制系统研究

项目名称： 融合视觉多信息的网络化控制系统研究

项目编号： No.61473182

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 自动化技术、计算机技术

项目作者： 杜大军

作者单位： 上海大学

项目金额： 81万元

中文摘要： 针对实际工业控制中视觉多信息(视觉感知的图像、非视觉感知和控制信息)给网络传输带来巨大压力，可能造成视觉多信息不完整，使得图像信息处理、控制系统分析与学习优化更加复杂的现实，本项目拟开展融合视觉多信息的网络化控制系统理论和实验研究：分析视觉多信息的传输机制和视觉图像特征，建立融合视觉多信息和网络参量的网络控制系统综合模型；从减轻通信负担的角度研究融合视觉多信息自适应触发机制和信息调度策略；综合分析网络通信约束下的视觉图像、信息传输对系统稳定性和鲁棒性影响的机理，探索图像特征参量、网络性能参量、触发器、调度与控制参数之间的关系以及系统性能评价体系；考虑图像压缩比等因素，设计基于反馈机制的控制器及补偿策略；研究图像特征参量、控制器结构和参数的自适应学习策略并分析收敛性，构建能测试、验证融合视觉多信息的网络化控制系统实验平台。最终，为网络化控制系统进一步的理论发展和实际应用奠定基础。

中文关键词： 网络化控制；视觉感知；稳定性；自适应学习

英文摘要： A large amount of visual information has brought huge pressure on network transmission during the process of industrial control, which not only causes lost images and relevant information, but also makes image process, control system analysis and learning optimization more complex than they should be. In this proposal, we intend to launch a study to networked control systems fusing visual images and other information. The transmission mechanism of visual images and other information is analyzed, and the features of visual images are extracted. A comprehensive model will be established. Then, an adaptive trigger mechanism fusing visual images and other information is studied, and information scheduling is developed to reduce the network load. We also analyze how visual images, information transmission and networked communication constraints interact, e.g. system stability and robustness. The relationship among the feature parameters of visual image, network performance paramters, trigger paramter, schedulng parameter and control paramter will be analyzed, and the system performance index is also investigated. Furthermore, the controller design based on the feedback mechanism and compensation strategies is studied. Moreover, the adaptive learning strategies for optimal control are investigated and the convergence properties are also investigated. Finally, an experimental platform based on visual and other sensors will be built for system validation. Eventually, the achievements of this project will lay a solid foundation for both theoretical study and practical applications of networked control systems.

英文关键词： networked control;visual perception;stability;adaptive learning