Strategy Research On Optimization Of Reactive Power Compensation In Low Voltage Distribution Networks

As electrical power supplier, low voltage distribution networks(LVDNs) have a large number of complicated electrical power users. Due to the limited efficiency of planning, operating, management and supervision, power loss of LVDNs remains unsatisfactorily high, and the low voltage problem is significant. Reactive power optimization has been put on distribution networks' planning agenda as a key technique for loss reduction and voltage improvement. In traditional distribution networks, reactive power optimization mainly aims at shunt capacitors(SCs) optimal allocation in some single load level, which apparently unable to fulfill the requirement of load with time-varying characteristic. Moreover, once local dispersed reactive power optimization devices have been deployed, some superfluous reactive power capacity of soft load buses may not be able to be transferred to the heavy load buses that short of reactive power, thus the efficiency of SCs reduced, neither the increasing load demand could be covered. Especially, with the integration of high penetration of distributed photovoltaic power generation(DPPG), resulting in single source uncertainty systems transited to multi-source uncertainty systems, which aggravates the difficulty of LVDNs reactive power optimization. Three-phase power flow calculation of distribution networks, as the basis of reactive power optimization, has a considerable development, but its convergence performance can't be guaranteed due to the impacts of duplicated neutral grounding of LVDNs. Collaborative optimization of LVDNs and SCs contains DPPG is a multi-extremum complex combinatorial optimization problem of high-dimension, nonlinear and non-differentiable, which results in the invalidation of traditional mathematical programming methods. Accordingly, on the criterion of given priority to efficiency, a mathematical model of reactive power optimization with the goal of minimizing the comprehensive costs over time with the consideration of costs of investment, power loss and voltage stability factor(VSF) has been derived, which is capable to reflect the operation reality of LVDNs and possess intuitive economical characteristics. The accuracy of the reactive power optimization model was supported by serval improvements as can be seen below.Above all, iterative three-phase power flow methods for distribution networks are susceptible to the duplicated neutral grounding of LVDNs, which may result in a poor performance in astringency. A linear expression of nodal voltage namely the non-iterative linear power flow has been presented by linearize the nonlinear terms of nonlinear formula consists of nodal load injection current and nodal voltage. Additionally, a compacted form of ? Fortescue equivalent nodal admittance matrix has been presented by applying Fortescue transformation to remedy the defects of zero-padding for the missing phase of phase-coordinate nodal admittance matrix. IEEE13 and 123-bus test system verify the effectiveness of improvements above.Secondly, for the collaborative optimization of LVDNs and SCs contains DPPG solved unsuccessfully by the traditional mathematical programming methods, and the defect caused by particle swarm optimization(PSO) vulnerability to falling into local optima, a new approach applying PSO with aging and challenging mechanism(ACM-PSO) has been introduced. In ACM-PSO,the mechanism of aging and challenging is adopted,and strategies of leader lifespan adjustment,challenger generation and its leadership assessment are designed to improve the convergence of PSO.The superiority of the ACM-PSO was verified through the application of IEEE16, IEEE33 and PG&E69 node test system.Last but not the end, different from the centralized reactive power compensation of substation, a novel reactive power compensation mode has been proposed, which centralized supply of reactive power capacity needed by reactive power compensation points in low voltage sides, namely the centralized compensation of reactive power in load sides. The proposed method has the function to distribute the capacity of reactive power according to the demand of reactive loads, which improved the efficiency of devices and the increasing load demand could be meet as well.In the end, the feasibility and accuracy of these improvements were validated through a plurality of typical IEEE test systems, thus, certain reference value for the refinement of distribution networks' planning can be provided.