分布式多源传感耦合的动态网络化系统状态估计研究

The state estimation for distributed systems based on multi-source sensors coupling is a hotspot and difficulty problem in the fields of intelligent manufacturing. The transmission uncertainty of dynamic networked system brings the complexity of system modeling and state estimation. The traditional state estimation for target mainly considers the perceptual information from single attribute sensor, and it is difficult to meet the practical necessity of the measurement precision. On the basis of the previous modeling and filtering for stochastic uncertain systems with related noise and communication constraints, this project orients to the system structure of multi-source sensing coupling, and studies the distributed information fusion method, and improves the data precision by using the information redundancy of real-time measurement systems. The project mainly research the following contents: 1) analyzing characteristics of multi-source sensors information, and considering the system for multiple spatial and temporal characteristics as well as coupling correlation, a distributed system model with information collaboration is established; 2) aiming at the multi-noise interference in the process of measuring information transmission, as well as the multi-communication constraints induced by the network, a re-organization method for measurement information based on event-trigger rule is established; 3) designing the minimum error covariance of state estimation based on event-trigger is proposed, and on the basis of covariance intersection, an optimal weighted fusion method is presented to improve the state estimation accuracy of systems. This project explores the key issues of distributed state fusion and estimation with communication constraints. The results can provide the novel theoretical, and method basis for ensuring effective information acquisition in distributed systems.

多源传感耦合的分布式系统状态估计是智能制造领域研究的热点与难点。动态网络化系统的传输不确定性带来系统建模和状态估计复杂，传统被测目标状态估计方法主要基于单一属性传感器感知信息，难以满足量测精度的实际需求。本项目基于前期研究含有相关噪声和通信约束的不确定系统建模和滤波方法，面向多源传感耦合的系统结构，研究分布式信息融合方法，利用实时量测系统的信息冗余度提高数据精度。主要研究内容如下：1) 分析多源传感信息感知特性，考虑系统的多时空特性与耦合关联性，构建分布式信息协同系统模型； 2) 针对量测信息传输过程中的多噪声干扰，以及网络诱导的多通信约束，建立基于触发规则的量测信息重组方法；3) 设计基于事件触发的最小状态估计误差协方差方法，研究基于协方差交叉的最优加权融合策略，提高系统的状态估计精度。本项目探索受通信约束的分布式融合与估计的关键问题，为保障分布式环境下的有效信息获取提供新的理论和方法。

复杂大规模网络化系统广泛地应用在能源、交通、物流等领域，由于其在现实应用中的强非线性和高度耦合等新出现的问题，极大地挑战了当前的计算工具。不同应用环境下具有复杂动态的多源传感耦合的分布式系统其状态估计问题，已成为分布式系统相关研究的核心问题之一。本项目主要研究了噪声和带宽约束下多源传感耦合的分布式系统的状态估计问题。应用线性滤波基本理论，探究具有降低不确定参数干扰的离线和在线状态信息辨识方法的基本特征，并应用在线状态特征辨识的最新成果，建立有效信息在线感知可靠性的理论方法。研究首先针对多时空尺度上包含误差、缺失、错误等不健全数据，借助离线数据关联状态特征提取方法获取可靠离线信息，修正感知信息并提高离线感知信息可靠性；借助实时在线数据，依靠在线状态特征辨识、相关性分析等手段，减少和消除冗余信息，获取有效信息并提高在线感知数据的可靠性。其次，借助线性滤波理论，考虑通信制约因素，研究具有容忍通信能力的离线和在线状态信息辨识和滤波方法，最终获取及时、可靠的系统各类节点运行状态。为多源传感量测系统自动排除随机干扰所引起的错误信息提供理论依据，并积极寻求提高数据精度的系统运行状态估计方法。. 基于前期研究，项目团队共发表科研论文12篇，其中SCI检索8篇，EI检索3篇。这些研究成果为分布式网络系统的协同状态估计的深入应用打下了坚实的理论基础。