Research On The Data-driven Control Of Complex Networks

Complex network control is a hot issue in complex network research in recent years.In the past research on complex network control,if there is a set of appropriate control signals that can drive the system to a specified final state in a limited time to control the entire network,the network is controllable,which is consistent with the controllability theory.The minimum number of driver nodes can be regarded as a control cost.Accurate network models are required to calculate the controllability of complex networks,but it is difficult to model large-scale complex network systems,and it is difficult to accurately build accurate models that meet the needs of network controllability research.Therefore,how to obtain the available data from the complex network system and realize the data-driven control of the complex network have become new methods in the research of the complex network control,which have strong practical significance.Through the system’s history and online data,the optimal control of complex networks is realized.Based on this concept,this paper does some work about the following research:First,a study of heterogeneous data reconstruction is carried out.Heterogeneous data refers to the collection of multiple sets of data at different times,different initial conditions and different final conditions obtained from the normal operating system for many times,which contain system information for each set of operational data.Secondly,based on the reconstructed heterogeneous data form,two data-driven control schemes for complex networks based on heterogeneous data are proposed.One is the control scheme obtained by the relationship between input and output data,and the other is the control scheme obtained by mining the system matrix.The minimum control energy is used as the output result to give the minimum amount of data required for heterogeneous data-driven control of complex networks.The feasibility of the two schemes is proved by experiments,and the reliability of the schemes under different proportions of input nodes is discussed.In addition,for the heterogeneous data disturbed by the independent and identically distributed ideal noise,a data-driven control scheme is proposed to solve the influence of the noise and realize the datadriven control of complex networks.Finally,the complex network data-driven control scheme proposed by Baggio is introduced into the multi-layer network,and its rationality is analyzed.By changing the singlelayer degree and interlayer density of multi-layer networks,several sets of experiments are carried out on ER-ER,WS-WS and BA-BA networks to verify that the minimum control energy obtained by this scheme has a consistent trend with the traditional scheme.Using the NewEngland power grid with 39 nodes to build a power-communication double-layer network,and research on the control of the network,the phase angle and frequency of the generator can be stably and quickly controlled to a synchronous state,and the input frequency is less affected.The control scheme is still effective when the normally distributed noise is disturbed.