Research On Smart Distribution Networks Dispatch Considering The Uncertainty Of Renewable Energy Generation

New energy generation (NEG) such as wind power generation and photovoltaicpower generation has provided fundmental support for enery saving and emissionreduction due to its clean, low-carbon and renewable propery. It is an effectiveapproach to ulitize NEG by intergrating it into power grids in a small and distributedmode. On one hand, it exerts positive effects, improving the economic operation andcuting down emissions. On the other hand, however, in some cases, network activepower losses could be increased and security problems such as overvoltage problemscould be caused, limiting the usage of NEG. Thus, it is necessary to apply reasonabledispatch strategy to reduce such negative effects.Therefore, tracking the reseach demand in distributed generation dispatch insmart distribution networks and considering the probabilistic and intervalcharacteristic of renewable energy generation, this thesis focuses on the stochasticdispatch, interval distpatch and the hybrid dispatch for smart distribution networks asfollows:A chance constrained optimization based multi-objective stochastic dispatchapproach is proposed for distribution networks. In this approach, the normaldistribution of forecated wind power is considered, and expected active power lossesand the risk level with respect to N-1contigenecies are considered as the objectives.The line power flow constraints, bus voltage constraints, swing bus power constraintsare satisfied with a predefined probability level. Therefore, for operators, a balancebetween economic operation and security can be made by propoer settings ofpredefined probability level. The two-point estimate method based probalistic powerflow and a multi-objective group search optimizer are used to solve the proposedmultiobjective stochastic dispatch problem.An operational robustness based distribution networks stochastic dispatchmethod is proposed. In this method, a modified two point estimate method is used toto guantee a high level of operational immunity of schedule to uncertainties inrenewable energy generation (REG). That means it is ensured as far as possible thatnon-violation of security contraints could be caused by the variation of REGand thusthe schedule has a high robustness. In addition, the ratio between standard deviationand mean value of network active power losses must be smaller than a given threshold,thus, the range of network active power losses can be controlled by the operators.A two-stage stochastic programming based distribution networks stochastic dispatch is proposed to fill the gap that economic and secure influcences of thedifference between the scheduled wind power and actual wind power can not beconsidered in the above approaches. In this approach, the distribution networksdispatch is divided into two stages: the first stage and the second stage. In the firststage, a wind power schedule is made before wind power is observed. In the secondstage, with the actual wind power, the recourse cost including curtailment cost,reserve cost and operational risk due to the difference between the dispatched andactual values of wind power is evaluated and considered with the dispatch cost as anobjective for the final determination of schedule. Simualtion results show theproposed approach can improve the security level at a small increase in cost, andsignificantly reduce the number of determistic and near violations of securityconstriants.A hybrid chance constrained optimization and operational robustnessapproach isproposed to consider normally distributed load and interval of REG. First, explicitformuations of chance constraints of swing bus active power and reactive power andbus voltage together with robust constraints of branch power flow are proposed basedon normally distributed load and interval of REG. Then, based on the proposed chanceconstraints and robust constraints, an approach aiming at optimizing the networkactive power losses with respect to forecasted REG and bus load is proposed.Therefore, the network active power losses in the forecasted scenario can beoptimized while the robust branch power flow can be achieved and the bus voltagesand swing bus powers can be satisified with a predefined probability level.An interval optimization based distribution networks interval dispatch isproposed to consider interval of bus load, REG and operational parameteruncertainties such as branch current limits. In this approach, the interval of networkactive power losses are optimized, and the constraints such as branch currents, busvoltages, and powers at the grid supply point are formulated as interval constriants,thus reflecting the problems such as lack of data and inaccurate parameters. Theinterval dispatch problem is transformed into a deterministic dispatch problem and theinterval power flow and improve group search optimizer are used to sovle theproblem.Focusing on the hour-ahead uncertain dispatch of smart distribution networks,this thesis proposes some approaches that can be used in different uncertaininformation scenario of REG, e.g. probabilistic information and interval information.The proposed approaches have a wide application prospect in the filed of hour-ahead uncertain dispatch of smart distribution networks...