Application Research Of Complex Network Theory On Time Series Analysis Of Near-surface Wind Field

The near-surface wind is the main carrier of gas diffusion,affected by several factors such as temperature,air pressure,buildings and complex topography,near-surface wind field exhibits strong nonlinear,non-stationary and uncertainty,which brought great difficulties to the estimation of poisonous gas diffusion trajectory and the locating of leakage source.Hence,deeply uncovering the underlying dynamic mechanism and the temporal-spatial evolution of near-surface wind field will contribute to a better understanding of gas diffusion behavior and will further provide valuable clues for the research on source locating and tracking of toxic gas leakage.In recent years,complex network theory has provided us a powerful framework for investigating nonlinear complex systems,especially those who are difficult to be described by accurate theoretical models,and has been widely used in the research of diverse research fields such as biological systems and meteorological systems.Hence,based on complex network theory,we investigate the optimal selection problem of high frequency anemometers,and thoroughly analyze the variation rules,the spatial correlation and difference of wind speed time series in the near-surface wind field.The main work of this paper is as follows:Firstly,aiming at the optimal selection of ultrasonic anemometers in different wind field environments,we utilize limited penetrable visibility graph algorithm with good anti-noise performance to analyze wind speed time series with the sampling frequency of 1Hz,2Hz,4.2Hz,5Hz,6.25 Hz,8.3Hz,12.5Hz,16.7Hz,25 Hz,50Hz in indoor and outdoor environments,respectively.Results of network characteristic parameters show that considering the price factor and the integrity of wind field information,it is recommended that the minimum threshold of the sampling frequency of ultrasonic anemometers in indoor environment is 8Hz,while the sampling frequency is at least 5Hz in outdoor environment.Secondly,aiming at the exiting defect that the mapping relationship of undirected complex network is too rough,this paper represents a novel method of constructing an improved weighted complex network based on visible angle measurement.The analysis results of the Logistics chaotic system and the Lorenz dynamical system with different noise intensities indicate that the proposed method has excellent sensitivity to subtle changes of signal.Finally,we use the proposed method to analyze the wind speed signal collected at different heights.Results indicate that the method can not only effectively distinguish the multi-point wind speed signal in the vertical direction of wind field,but also accurately reflect the spatial arrangement law between the wind speed signals.Thirdly,the multivariate complex network approach combining with multi-scale analysis is adopted to deeply uncover the complex dynamic behaviors underlying the multi-scale fluctuation structures of 3D wind field.Compared with classical univariate time series method,the approach can accurately distinguish the 3D wind speed signals collected from indoor and outdoor environments,and can better quantify the essential difference of wind field at different time scales.