基于云计算的复杂网络化多智能体预测控制研究

With the rapid development of Internet of Things technology, networked multiagent systems become more and more complex. This proposed project mainly studies complex networked multi-agent predictive control via cloud computing. Considering complex dynamics, communication constrains, internal uncertainties external disturbances of systems, an integrated mechanism-data model of complex networked multiagent system is established. To compensate for the complex communication constraints actively, reduce the model uncertainties and external disturbances effectively, and achieve the desired control performance completely, a complex networked multi-agent predictive control scheme is proposed. To solve the distributed computing problem of complex networked multiagent predictive control systems, a complex networked multiagent predictive control method based on cloud computing is presented, an efficient implementation control algorithm is designed, coordinative predictive control and optimal predictive control of complex networked multi-agent systems via cloud computing are studied. To guarantee the stability and consensus of networked multiagent systems at the same time, the simultaneous stability and consensus analysis method of networked multiagent control systems is proposed, and simultaneous stability and consensus criteria are derived. The effectiveness of the proposed methods in this project is demonstrated through many real-time simulations and practical experiments. It will lay a solid foundation for the theory and application research of complex networked multiagent cloud predictive control.

随着物联网技术的迅猛发展, 网络化多智能体系统变得越来越复杂，本项目主要研究基于云计算的复杂网络化多智能体预测控制。针对系统的复杂动力学特性、通信受限、内外不确定性，建立复杂网络化多智能体系统的机理-数据综合模型。为了主动补偿复杂的通信受限，有效地抑制模型不确定性和外部干扰，完全达到期望的控制性能，将提出复杂网络化多智能体预测控制策略。为了解决复杂网络化多智能体预测控制的分布式计算问题，提出基于云计算的复杂网络化多智能体预测控制方法，设计高效的多智能体云预测控制实现算法，研究复杂网络化多智能体云预测协调控制和复杂网络化多智能体云预测优化控制。为了同时确保网络化多智能体系统的稳定性和一致性，提出网络化多智能体控制系统稳定性和一致性同步分析方法，获得稳定性和一致性同步判据。通过实时仿真和实物实验，验证本项目所提方法的有效性。为复杂网络化多智能体云预测控制的研究奠定一个理论和应用基础。

随着控制系统技术、通信网络技术和云计算技术的发展，基于云计算的网络化控制系统呈现出强劲的发展趋势，已应用于物联网和工业互联网等领域。本项目将云计算和预测控制以一种通用的形式引入到复杂网络化多智能体控制系统，以克服其具有挑战性的问题，尤其是传统网络化多智能体控制系统中存在的实时大数据、通信延迟、大计算量和多任务协调等问题。鉴于云计算的优点，本项目提出了一种基于云计算的复杂网络化多智能体预测控制策略。研究了基于云计算的网络化多智能体预测控制，网络化非线性多智能体系统的数据驱动预测控制，通信受限的复杂网络化多智能体系统的预测控制，网络化多智能体系统的鲁棒预测控制，网络化多智能体系统的学习预测控制，网络系统的递推状态估计。深入分析了复杂网络化多智能体系统的一致性和稳定性，获得了多种网络化多智能体预测控制的一致性和稳定性判据。开发了云计算的复杂网络化多智能体预测控制实验平台，进行了大量的复杂网络化多智能体预测控制仿真和实验，成功地验证了研发的各种复杂网络化多智能体预测控制方法。为更进一步研究复杂网络化多智能体控制系统，奠定了一个坚实的理论基础。