Distribution Network Reactive Power Planning Based On Sensitivity Analysis And Evolution Strategy

Power system is developing rapidly in recent decades, the scale of power gridis increasingly larger, the structure of power grid is increasingly complex, the gridrun modes is also becoming more and more complex. However, compared with asurge of social demand and the rapid development of distributed power supply andmicro power grid, the speed of distribution network construction is relatively slow,the phenomenon of the voltage quality is low caused by reactive power deviceconfiguration is not reasonable, the rate of operation is not high, and problems of ahigh network loss are existing for a long time. Predictably, surrounding with theconstruction of distribution network, these issues, such as the selection ofcompensation position of reactive compensation equipment and the configuration ofcapacity of reactive power equipment will more and more be taken seriously.Finding the right position of reactive power compensation is the first step ofthe reactive power planning. At present, a lot of theories are applied to powersystem used in compensation point selection, including the sensitivity analysis,modal analysis and the definition of load impedance torque, etc..The method ofsensitivity analysis is introduced and applied to select compensation posi tion in thispaper. Considering the time variability of the power system load, introducing theMonte Carlo method to sensitivity analysis to stochastic simulation, through thesimulation experiments of limited time, in the end by comparing the probability ofeach node chosen for the compensation point in limited time experiments todetermine the reactive power compensation position. Finally taking the IEEE-34system for example analysis to specify the selection process of reactive powercompensation points.After selecting the compensation position, using the Evolution Strategy for thedetermination of compensation capacity in this paper. According to thecharacteristics of distribution network, at the same time, in order to better play thecharacteristics of Evolution Strategy, some appropriate improvements are made tothe evolution strategy methodin in this paper. In the formation of the initialpopulation, introducing the multidimensional space from mathematics and usingEuclidean distance to determine the similarity of initial individuals. In the link ofmutation, the different populations use the different mutation strategies. In the elitepopulation using cauchy mutation instead of gaussian mutation, broadening thespace of solution, increase the probability of finding the optimal solution. Thispaper construct the fitness function of the adaptive weight coefficient, and usingtwo different choose strategies to the two populations, it will have a larger population diversity using a small population size. Introducing the concept of chaossearch to Evolutionary Strategy, increasing the probability of searching the optimalsolution.Finally taking a actual grid case for example calculation, verifying thefeasibility of the proposed algorithm, and by comparing with the unimprovedalgorithm, verifing the improved Evolution Strategy have better global searchingability.