Research On Optimization Strategies Of Load Balancing In Electric Power Communication Network

The profound changes in economy and energy field at home and abroad havebrought new opportunities and challenges to the development of electric powercommunication network. In recent years, with the continuous construction of smartgrid, modern grid’s features of information, automation and interaction are becomingincreasingly apparent, and the power communication network, which plays a strongsupporting role for the grid, is also undergoing changes. With the deep penetration ofinformation communication technology to electric power production and management,the power grid is becoming increasingly dependent on communication networks. Thereliability and transmission performance of the communication network will play akey role for the security and stability of the grid.With the increasing complexity of network topology and the gradually enrichedbusiness demands, the load imbalance problems in electric power communicationnetwork become more serious. Certain core nodes or links carry a large amount ofbusiness data, which increases the vulnerability of the network, and also reduces theutilization efficiency of network resource. Traditional optimization strategies didn’tconsider the topology characteristics and business flow attributes in electric powercommunication network, so they were always difficult to use directly. Based ontopological characteristics and flow attributes of electric power communicationnetwork, this thesis studied the problems of topology optimization, flow analysis andprediction and routing optimization in electric power communication network. Themajor contribution of this dissertation is specifically stated as follows:(1)Based on the robustness and vulnerability analysis, one of topologyoptimization algorithm based on the matrix of node betweenness plus was proposed,which is used for topology optimization. The optimization algorithm firstly generateda set of optimized links based on matrix of node betweenness plus, and then achievedoptimization goals by adding a limited number of links. It also used evasivemechanisms to avoid excessive concentration of network traffic. The results ofexperiment show that the algorithm could effectively enhance the network’s ability towithstand deliberate attack and delay the first-performance dips in the network. It canalso improve the bandwidth utilization efficiency and reduce the probability ofcongestion of the key nodes. Hence the optimization algorithm is more targeted andhas more effective results for the power communication network topology. (2)According to the flow characteristics of power communication network, anARIMA (Autoregressive Integrated Moving Average model) was established forelectric power communication flow analysis and forecasting. According to theprocedures and methods of ARIMA model, the research was focused on the videobusiness and production data traffic analysis and forecast. Through data preprocessing,model parameters selecting and mode building，the power communication trafficanalysis and forecasting model was founded. The residual test method was used toverify the applicability of the model. The simulation results indicate that the ARIMAmodel fully considered the flow attributes of ups and downs, trend and cyclical. Itcould better adapt to power communication traffic characteristics, and effectivelyimprove the accuracy of model fitting and prediction accuracy.(3)A VECMP (Variable Equal-Cost Multi-Path) model based on link-weightcontrolling was posed to achieve balanced load of electric network. The improvedgenetic algorithm IGA-VECMP (Improved Genetic Algorithm for VariableEqual-Cost Multi-Path) was used to solve the model. The VECMP model was basedon the principle of link weight variable. It introduced controllable parameters Sktothe ECMP (Equal-Cost Multi-Path) model. According to the proportion of nodeconfiguration, the nodes with high configuration priority will have the ECMP functionby dynamic selection strategy. Then the network load will be distributed more evenly.In order to solve the VECMP model, an improved genetic algorithm IGA-VECMPwas proposed. Through the optimization of selection, crossover and aberrance, theefficiency of traditional genetic algorithm was enhanced. An improved algorithm wasused to calculate the variable equal-cost multi-path problems involved in the fitnessfunction. Simulation results showed that the VECMP model and solution algorithmcan quickly obtain the optimal weight of the network configuration and effectivelyachieve the network load balancing. And it significantly reduced the maximum linkutilization in the network and enhanced the efficiency of the network bandwidthutilization. The model and solution algorithm provided extensive value in routingoptimization of electric power communication network.