Research And Application Of Reactive Power Optimization Of Distribution Network Based On Improved Simulated Annealing Particle Swarm Algorithm

With the rapid growth of China’s economy,the demand for electricity from all walks of life is also increasing rapidly.At present,a large proportion of the rural population began to urban influx,such a situation makes the power supply in the city gradually increased,the overall level of electricity load in the city in an exponential growth,the increase in the load of the power grid in the city will also make the distribution network structure becomes more and more complex,including the grid connection will also become more and more complex.Nowadays,the power industry is gradually approaching to the market-oriented economy,how to further improve the power quality and optimize the resource allocation,to reduce the line loss to a new level,and to improve the economic effect of the power network is the top priority of the current power research.According to the real demand,a large number of researchers keep improving the existing artificial intelligence algorithms to make them more effective for reactive power optimization of distribution networks,to overcome more or less deficiencies of existing algorithms,and to improve the power quality.In this paper,the objective function is to minimize the active network loss in the network and construct a mathematical model for the reactive power optimization problem of the distribution network,and the main algorithm is a simulated annealing particle swarm algorithm,on the basis of which two improvement strategies are proposed,namely,improving the inertial adaptive weight and introducing the adjustment factor to constrain the learning particles,which can make the algorithm have stronger search capability.The improved algorithm can realize the self-adjustment of inertia weights,which makes the algorithm obtain stronger overall search ability in the early stage of operation and stronger local search ability in the late stage of operation.Speeding up the whole process makes the reactive optimization process more efficient.Invoking the adjustment factor to constrain the learning particles makes the final computed results apply the Metropolis criterion to jump out of the local area and still have the ideal convergence characteristics,which not only solves the trouble of inaccurate local optimal solutions,but also improves the convergence of the results.Next,the improved simulated annealing particle swarm algorithm is compared with the pre-improvement algorithm through four different functions,and the results show an improvement in convergence speed and accuracy.The compensation nodes are identified for reactive power compensation according to the sensitivity calculation.In this paper,the IEEE-33 standard node system is used as the target system network for reactive power optimization verification,and the results are compared with Matlab.The improved simulated annealing particle swarm algorithm can optimize both the average node voltage quality and active network loss compared with the pre-improvement algorithm,which can improve the former by 2.63% and reduce the latter by 7%;when applied to an actual line in Panjin City,the improved algorithm can improve the average node voltage quality by 1.67% compared with the pre-improvement algorithm,and for Active network loss is reduced by 12.79%.The improved algorithm has better convergence accuracy and rate than the previous algorithm,and the results show that the application of the improved simulated annealing particle swarm algorithm can improve the voltage quality and active power loss of the distribution network and solve the existing problems of Panjin distribution network by controlling the number of shunt capacitors in accordance with the sensitivity calculation and achieving reactive power compensation.