| The prediction of icing load of transmission line plays an important role in the practice and economy of power network planning,in particular,the analysis and prediction of ice-covered disasters has become a very important research direction.Affected by the ice space scale of complex power system,local geography is very complicated,cause ice load fluctuation is bigger,change law of electric power system with the ice load factors and complicated changes,many related to power grid operation,including planning and seheduling,load balancing and load follow,are performed in different granularity of space and time.According to different geographical environment and climate environment,combine the influence factors and law of overhead transmission lines ice,respectively from the Angle of time granularity,establish proper overhead transmission lines ice load forecasting system,and power transmission and transformation is given according to the results of the prediction system of early warning,in the end for power transmission and transformation system of deicing and maintenance decision service is urgently needed to solve problems.Based on the data-driven method,this paper aims to solve the problem of insufficient attention to multi-granularity of space and time of the warning model existing in the current transmission line icing prediction model A spatial-temporal multi-granularity situation assessment model for ice cover warning for overhead transmission lines is proposed.The main research results are as follows:1)For the dynamic,nonlinear,uncertain and sudden characteristics of the time series of icing load,using the time multi-particle signal analysis method wavelet analysis and Empirical Mode Decomposition(Empirical Mode Decomposition,EMD).The empirical modal decomposition(EEMD)is used to analyze the time series of icing load on transmission lines.Through the example,this model can be overhead transmission lines ice load as non-stationary signal decomposition in the remaining components of different level,the trend component and multiple periodic components with different properties of signal,reduces the original ice load time series is non-stationary and sensitivity on the result of forecast,in order to achieve the purpose of better track field ice change rule.2)In view of the traditional time series prediction model does not consider multiple time granularity under poor prediction accuracy and prediction results of single faults,based on multi-scale line ice time series analysis on the characteristics of the different components,on the basis of the combination is suitable for processing temporal data cycle RVM and depth of the neural network model,puts forward a new transmission lines ice load time series combination forecast model.The example verifies that EEMD-RVM and EEMD-RNN methods are proposed to predict the icing load of multi-time granularity lines,and SVM,Elman and EEMD-Elman comparison models are proposed for comparison.The method has higher prediction accuracy and good application value.3)In view of the lack of overhead transmission lines ice cover is a lack of comprehensive three-dimensional monitoring and warning and multi-dimensional three-dimensional monitoring and early warning model of the problem,based on geographic information associated with spatial position data of the earth’s surface,and the transmission line prediction of ice load time granularity,establishing transmission lines ice warning multi-granularity situation assessment system,breakthrough and implements the point focus on site monitoring and early warning in Yunnan region in the transmission line complex channel environment,the key line linear monitoring and warning and key technologies,such as surface type wide area monitoring and early warning,the running state of the transmission line is established multidimensional stereo monitoring system,combined the technology of data fusion,storage and analysis,It provides data support for the operation and maintenance of Yunnan regional transmission line icing. |
PDF Full Text Request