Complex network node classification method based on graph attention network

Abstract

The invention discloses a complex network node classification method based on a graph neural network so as to solve a difficult problem of complex electromechanical system coupling network community division. The method comprises the steps: applying detrending coupling correlation analysis to the calculation of the correlation between monitoring variable nodes, and carrying out the primary screening of correlation coefficients through the introduction of Gaussian noise; introducing a scale index, and carrying out secondary screening on the correlation coefficients; taking the monitoring variable as a network node, converting the correlation coefficients into edge connection weights, and constructing an undirected weighted complex network; starting from a static community detection algorithm of global modularity optimization based on module gain, taking each node in the network as a partition, calculating modularity gain of a neighbor node to the current community according to a modularity function, judging community attribution of the node according to the modularity gain, and obtaining primary partition of the network node; taking the community primarily divided by the network asa node again, carrying out a new round of iteration on a new network, obtaining an optimal community division result of the network when the modularity reaches a maximum value, and taking the result as an initial training label of the graph attention neural network; carrying out training based on real-time monitoring data through a graph attention neural network, realizing node classification of the complex network, thereby providing a reliable basis for accurate description of the complex electromechanical system coupling network.

Description

technical field [0001] The invention relates to the technical field of complex electromechanical system coupling network community division, in particular to a complex network node classification method based on graph attention neural network. Background technique [0002] There are many monitoring variables in the complex electromechanical system of the process industry, and the material flow and energy flow in the production equipment and pipelines are closely coupled with the control and information flow in the communication network. Taking the compressor unit of the typical production equipment of a chemical enterprise as an example, the layout of the system is as follows: Thousands of monitoring points including pressure, temperature, flow, liquid level, vibration, speed, switches and alarm signals, etc., and the mutual coupling of monitoring variables in the system essentially form a dynamic The changing network diagram, and the complex network has become a powerful to...

AI Technical Summary

Problems solved by technology

[0003] In order to solve the above-mentioned problems existing in the prior art, the purpose of the present invention is to provide a complex network node classification method based on the graph attention network, aiming at the difficult problems of complex electromechanical system modeling and community division, the application of detrending coupling fluctuation analysis method to calculate Correlation coefficient, establish a data-driven complex network model, and realize complex network node classification based on graph attention network, and complete community division

Images