Research On Distributed Energy Management And Optimal Dispatching Methods For Smart Grid With Wind Power

As the "nerve center" of smart grid,energy management system realizes dispatching and decision-making and controlling of distributed energy in the power grid,which is the key to ensure the safe,economic,efficient and stable operation of smart grid.Traditionally,centralized strategies have been applied to energy management.These centralized strategies require a central dispatcher to collect global information from all committed generators,and then to make dispatching decisions.However,as a large amount of distributed energy resources is penetration into power grid,centralized energy management imposes a greater communication burden on central dispatcher,and is more likely to be disabled when the smart grid suffers extreme events such as natural incidents.In contrast,distributed energy management strategy manages smart grid just by exchanging local information among neighbors,and thus can protect privacy of the committed units and is more scalable for future expansions under circumstance of the penetration of distributed generations,and is more resilient to single-point failures due to extreme events.Therefore,distributed strategy will become principal energy management method to smart grid with integration of renewable distributed energies in the further.As an important distributed renewable resource,wind energy has inherent advantages over traditional fossil fuel energy due to its environment friendliness.With the increasing penetration of stochastic wind power into smart grid,there is a pressing need for considering energy management involving wind power.However,the characteristic of intermittent and stochasticity of wind speed leads to uncertainty in wind power output,which causes the design of distributed energy management strategies more difficult.Hence,this paper proposes distributed strategies for energy management with wind power,which is expected to provide decision support for power system decision makers in energy management of smart grid under the penetration of wind power.The specific contributions include the following three aspects:(1)Energy management for smart grid with stochastic wind power is investigated,where the ramp rate limits of the thermal generators(TGs)are taken into consideration and the wind power profile during different operation intervals is characterized by different Weibull distributions.The parameters in each dispatch interval are estimated by maximum likelihood method according to wind speed data.Combined with distributed gradient algorithm,the distributed computation procedure by Alternating Direction Method of Multipliers(ADMM)is deduced.The convergence speed of the proposed algorithm is accelerated by Armijo line search algorithm,which is introduced to obtain step size adaptively in the iterations.Most importantly,compered with the existing distributed projected gradient method(DPGM)for wind power energy management,the proposed algorithm exchanges the local variables rather than the vector of global estimated optimal values in each iteration,and thus has a better convergence performance.(2)Dynamic economic dispatch(DED)in energy management problem with wind power is studied,where the carbon emissions trading cost and its impact on the penetration of stochastic wind power is considered.Since the previously proposed consensus-based ADMM needs a proper algorithm initialization to meet the power balance constraint,a novel initialization-free distributed algorithm is put forward for the proposed DED model.By integrating dynamic average consensus algorithm into ADMM,our proposed algorithm can track the power balance mismatch,and hence has ability of adapting to time-varying load demands and plug-and-play operations without reinitialization.In addition,our proposed algorithm does not need to exist explicit expression of the inverse gradient function of objective function like that in existing initialization-free distributed algorithm.(3)Chance-constrained energy management with wind power is investigated,where spinning reserves with probability constraints provided by TGs are introduced to deal with the uncertainties,so as to respond to the deviation of the available and the scheduled wind power in a certain confidence level.By assume that the uncertainties of wind power are characterized by Weibull probability,the proposed problem is formulated as a stochastic programming model,and then is formulated into a deterministic model.Since the problem is difficult to be solved by the previously proposed distributed method due to the coupling of equality and inequality constraints,the original problem is reformulated into a novel compact form,where the decision vector of each committed generator is individual decision variables instead of the collection of all local decision variables.Based on this,a distributed tracking ADMM algorithm is proposed to solve this model with such a compact form.Analogously,this formulation is comprehensive to describe several energy management models,and is scalable to integrate more distributed generators.