Research On Network Topology Identification Based On Compressed Sensing

With the increasing complexity of power systems in engineering and other scientific fields,the study of interconnected systems or dynamic networks has become one of the important research topics in different fields such as smart grids,social networks,and transportation systems.Topological structure and dynamics are closely related in these complex networks.On the one hand,describing the topology of the network can reveal its structural characteristics.On the other hand,studying the dynamics of the network can reveal its functional characteristics.However,the study of network topology occupies an important position in complex systems.It is particularly important for understanding and controlling complex systems to understand the relationship between the network structure and various dynamic behaviors from the measurement data.In recent years,compressed sensing has been a research hotspot.It can recover original sparse signals accurately with less measurement data.Therefore,compressed sensing is often used for the study of network identification,which is an efficient network identification method compared to other identification methods.However,the networks that use compressed sensing to identify are mostly sparse networks,and the current research on non-sparse networks is still relatively small.At the same time,the influence of noise on identification is also not negligible,which can reduce the success rate of identification or increase measurement data.In addition,the linear network system is also a simple system that is often applied in various fields.However,the state of the nodes in the system is very stable and has a strong correlation,and the topology structure of the network cannot be identified using compressed sensing.To solve the above problems,this paper proposes a network topology structure identification method that combines compressive sensing and QR decomposition.The main research results and innovations of this paper are as follows:(1)A new method integrating QR decomposition and compressed sensing is proposed to solve the network topology identification under the assistance of the input noise.This paper uses noise to drive the state among the nodes in the linear network system to break the stability of the system state,but also make up for the defects of compressed sensing to identify the non-sparse network topology structure,and details some of the factors affecting the network topology identification.The analysis proved the feasibility of the proposed method.(2)In this paper,the identification method of network topology structure with integrating compressed sensing and QR decomposition is applied to the identification effect of different real networks and model networks.Aiming at these different networks,we compare the two identification methods of traditional compressed sensing and integrating compressed sensing-QR decomposition.Experimental results show that the proposed method is better than the traditional compressed sensing method.