Research On Target Distributed Observation Problem Based On Complex System

In actual engineering,there exist complex systems that occupy wide areas(such as the temperature control system of a large production workshop)or have a high dimension(such as a power grid).The collection and measurement of the output information of this kind of system is difficult to be realized by a single sensor.Instead,a sensor network composed of multiple sensors is needed,in which each sensor performs the measurement of its partial output information.Since part of output information cannot guarantee the observability or detectability of the system,classical observers cannot realize the state estimation of such systems.Therefore,distributed observers based on partial output information are introduced.Based on the existing achievements of distributed observers,this paper considers the distributed observer design problem under peer pressure and the distributed observer design problem of moving targets,as follows:(1)Aiming at the problem of distributed observer design for discrete complex target systems,the directed communication topology and the transformation-based d-path Laplacian matrix are first used to characterize the influence of indirect neighbors in the sensor network;then based on the detectability decomposition,a weighted coupling matrix is introduced to design a simple distributed full-dimensional observer algorithm;finally,in order to reduce the number of integrators in the distributed observer,a distributed reduced-dimensional observer is constructed.(2)For a Linear Time-Invariant(LTI)moving target system with constant speed,in the design of distributed observers,Using the local measurement,each sensor is endowed with two coupled dynamic algorithms: one is a distributed state estimation algorithm based on the consistency of targets,which constructs a distributed observer related to the sensor network using the constant coupling gain.The other is a leader-following clustering algorithm based on the target output to guide the movement of the sensors,allowing moving sensors to avoid collisions,maintain communication,and track targets.(3)For a LTI moving target system with time-varying speed,a distributed state estimation algorithm based on the consistency of targets is also used,and this paper uses node adaptive gains to design distributed observers.In order to solve the theoretical analysis difficulties caused by adaptive algorithms,a clever non-singular transformation based on Lyapunov stability theory is introduced.It is proved that a distributed observer can be formed under the conditions of target detectability and initial topological connectivity of the mobile sensor network.In addition,for the mobile control of the sensor network,this paper designs a new maneuver algorithm to track variable speed moving targets.