Research On Line Loss Calculation Method Of Distribution Network

The line loss calculation of distribution network is an important technical meansin power system which can be helpful to lower power loss. It is also the basis thatmakes line loss management scientific, standardized and institutionalized. Accurateand convenient methods of calculating line losses can help to formulate reason ableloss decreasing measures, improve electricity supply ability and increase theeconomic benefit of power enterprise.However, the structure of distribution network is complex, branch line isnumerous, accurate and complete line load data is seriously lacking. These reasonslead conventional line loss calculation methods to be difficult to implement. In orderto solve the problem, this paper takes advantage of the nonlinear processing ability ofneural network to research on methods for calculating the line losses of distributionnetwork which is easy to operate, has high feasibility and meet engineering precision.Firstly, according to the characteristics of medium voltage and low voltagedistribution network, the paper analyzes existing methods for line loss calculation inmedium voltage and low voltage distribution network and points out their applicablecondition and shortages.Secondly, in view of the problem that the structure of medium voltagedistribution network is complex, operation data is incomplete, conventional line losscalculation methods are difficult to implement, the Radial Basis Function (RBF)neural network is applied to calculate the line losses of medium voltage. Byestablishing the corresponding RBF neural network model, the method takesadvantage of its strong regression ability to map complex nonlinear relation betweenline loss and feature parameters of distribution net, and memorizes the rule of lineloss varying with distribution circuit structure and operation parameters.Thirdly, by improving Adaptive Second Mutation Differential Evolution(ASMDE) algorithm, the idea to restructure crossover probability factor andapproximate optimal preservation strategy have been put forward. Adopting improvedASMDE algorithm to optimize integrally the structure parameters of RBF neuralnetwork, the method overcomes the shortcomings that conventional network trainingalgorithm is easy to fall into local optimum and the hidden layer and output layerstructure parameters are determined separately. The model and algorithm have been verified through simulation example. The simulation results have proved thefeasibility and precursor of the proposed method of line loss calculation for mediumvoltage distribution network.Lastly, the methods for calculating line losses of low voltage distributionnetwork are studied. Low voltage distribution network has the characteristics thatpower supply is complicated, branch line is numerous, the elect ricity load along theline has no strict rules, the degree of automation is not high, parameters of line andload data are lacking seriously. Because of these, Back Propagation (BP) neuralnetwork has been used to calculate and analyse the line losses of low voltagedistribution network. At the same time, the input parameters of the BP neural netwo rkare analyzed in detail. Accordingly, finding out the main parameters that cause linelosses of low voltage distribution network changes and using of them as the inputparameters of BP neural network model. Using the neural network toolbox in Matlabto complete the network’s training. Through the simulation example, the resultsshowed the accuracy and practicability of the model which used to calculate linelosses of low voltage distribution network.