Research On Traffic Model In Communication Network For Smart Power Distribution And Consumption Systems

With the construction of the information smart grid, massive secondary equipment continually access to the intelligent power distribution and consumption systems, the demand for update of real-time data becoming increasingly important, the traffic in communication network for smart power distribution and consumption systems increasingly increase and complicated change. Communication Network for Smart Power Distribution and Consumption Systems are getting more and more complex, the accurate depict of date communication network has become an important problem to be solved. Therefore, create a traffic model that can accurate depict these character based on intensive analysis the character of the traffic in communication network for smart power distribution and consumption systems has become the primary research topic of realize the intelligentize of smart power distribution and consumption systems.Firstly, study the character of the traffic in communication network for smart power distribution and consumption systems, combine all kinds of quality of service (QOS) metric to class divide the traffic, analysis the transfer process for the traffic, establish the transaction flow diagram for every process, and analysis the character of the self-similar and multifractal for the traffic in communication network for smart power distribution and consumption systems.Secondly, an internet traffic model that gives consideration to both self-similar characteristic and multifractal characteristic is proposed in this paper. This model preserves Gaussian distribution of the independent wavelet model to produce multiplier, so the reconstructed traffic is able to representation the self-similar characteristic of the actual network traffic. In addition, the model is able to maintain the multifractal characteristic of the actual network traffic by revise the wavelet energy. This model can perfectly depict the characteristic of the actual network traffic, possess better fitting results.Finally, a traffic prediction model based on the correlation between wavelet coefficients is proposed in this paper. This model is fully taken into account of correlation between wavelet coefficients in the error function in the process of wavelet neural network training, and uses covariance to characterize this correlation. To modify the weights of neural network by gradient descent method using error function, then obtain prediction traffic through reconstruction the predicted value of all coefficients. This model is able to estimate more accurately of the actual traffic in communication network for smart power distribution and consumption systems compare with the wavelet neural network prediction model.