人机物融合感知网络的自主决策与闭环优化机制研究

With the rapid development of perception, storage and computing ability, the real-time sensing of the physical world has merged into every aspect of our daily lives. Smart sensing and its machine intelligence has become the current trend of science and technology. One of the cornerstone and also the biggest challenges is the intelligent decision-making based on sensing data. The goal of this study is to propose a new decision-making mechanism for human-cyber-physical system. This mechanism is based on a three-factor decision process model combining human, machine and the real world spaces. By knowledge discovery, the system can autonomously establish the decision-making rule and optimize the decision ability by a closed-loop iterative optimization mechanism. The innovation of this study is mainly embodied in the three key aspects: autonomy, human-factor and closed-loop. The expected results of this study can further improve the intelligent decision-making ability of the human-cyber-physical system, shifting the traditional two-factor decision mode into a more intelligent, closed-loop iterative, three-factor decision-making process.

由于感知、存储和计算能力的快速发展，对物理世界的实时感知已经深入到日常生产、生活的各个方面，感知机器智能已成为目前科技发展趋势，基于感知数据的决策是各类智能系统面临的共性关键问题之一。本研究的目标是针对人机物融合感知网络提出一种新的决策机制，基于“人”、“机”、“物”融合的三因子决策过程模型，通过自主发现感知网络中的决策知识，建立由人因驱动的决策规则学习能力，实现基于实时决策反馈的闭环迭代决策过程优化机制，支撑形成具有自主学习和演化能力的人机物融合感知决策系统。课题的具体研究内容包括（1）三因子闭环决策模型研究、（2）基于语义约束演算的自主决策知识发现方法研究、（3）闭环迭代人机物融合决策优化方法研究。本课题的创新性主要体现在“自主、人因、闭环”三个关键方面，预期成果能够进一步提高人机物融合系统的智能决策能力，由单一过程、双因子的传统决策模式向更加智能的闭环迭代、三因子决策过程演化。

项目完成了“针对人机物融合感知网络提出一种新的决策机制”这一主要研究目标，通过自主发现感知网络中的决策知识，建立由人因驱动的决策规则学习能力，实现基于实时决策反馈的闭环迭代决策过程优化机制。支撑形成了面向高铁列控系统的情景事件建模和检测技术、智能电器的情景指纹和和行为检测技术、大规模微服务系统中的异常情景感知和根因检测技术研究三方面的关键技术研究。项目研究成果已发表学术论文17篇，申请专利2项，获得软著权3项。