Multi-stage Robust Dispatch Of Power System Considering Uncertainties Of Renewable Energy

The traditional dispatching method is based on deterministic optimization and lacks an understanding of the uncertainty of renewable energy output.In the face of the access of a high proportion of renewable energy,the operator must adopt a conservative power generation plan,leading to an inefficiency of conventional units and curtailment of renewable energy.Based on the accurate modeling of the uncertainty of renewable energy output,this dissertion proposes a multistage robust dispatch method for the power system,which expands the optimization space of flexible resources such as energy storage and integrated energy.The resulting schedule is more economical and reasonable while meeting extreme scenarios,which is of great significance for enhancing operators' awareness of the uncertainty of dispatching plans and leveraging the flexibility of the power system itself and other energy systems.The main contribution includes five parts:(1)A data-driven method for constructing wind power forecast error uncertainty set is proposed.Via non-parametric data clustering method,the output prediction error data of multiple wind farms is grouped into several components,each characterized by a Gaussian distribution.The clustering result is used to generate the data-driven uncertainty set.Compared with the traditional polyhedron set,the data-driven uncertainty set can more accurately reflect the multimodal distribution characteristics of wind power output forecast errors.A data-driven robust unit commitment(DDRUC)model is established afterwards,and an algorithm based on constraints and column generation is proposed.The DDRUC reduces the conservativeness of dispatch plan and makes the unit commitment more reasonable.(2)Aiming at the intraday economic dispatch problem considering the uncertainty of renewable energy output,a multi-stage robust dispatch method is proposed.This method expands the time window of traditional look-ahead economic dispatch.By introducing the worst-case value function,the worst-case scenario of the cost of future periods caused by current decision can be measured.In this way,energy storage resources can be coordinated throughout the day to alleviate wind curtailment and load shedding.Fast Robust Dual Dynamic Programming(FRDDP)algorithm is proposed to efficiently solve the model.Compared with the existing algorithm,FRDDP significantly reduces the scale of the problem while maintaining reliable convergence of the algorithm.Considering the non-linear characteristics of charging and discharging of the energy storage system,a state transition equation based on estimationcorrection is proposed,which is suitable for multi-stage problems.The optimal scheduling strategy and extreme scenario obtained by this approximation method are close to the real situation.(3)The traditional spinning reserve is expaned to the concept of economic operation region(EOR),which suitable for thermal power units,energy storage and load aggregators.A two-stage robust scheduling method based on EOR is also proposed.The optimal solution coverage theorem of the EOR is proved,that is,the change of the dispatch plan caused by the randomness of renewable energy output will not exceed the EOR.Adding the worst-case value function of energy storage to the objective function of the second stage of the model further enhances the control strategy of energy storage.An algorithm based on nested Benders decomposition is designed to solve the proposed model.Compared with deterministic method,in scenarios with a high proportion of wind power and large forecast errors,the proposed model has significant advantages in the sense of wind power consumption and total cost.(4)Focused at the scenario where the district heating system participates in the power system regulation to improve the consumption of renewable energy,a decentralized and coordinated multi-stage robust dispatch model for electric-heat integrated system is proposed.This model protects the privacy of different energy operators while leveraging the flexibility of multiple energy supply systems.Tackling the complex structure of the model with multi-stage and multiagent,the worst-case spatial-temperal value function is proposed,so that the decision of each stage of each subsystem can take into account the impact on the rest of the time period and other subsystems.A decentralized robust dual dynamic programming algorithm is proposed to solve the model.The algorithm can converge in a short time for medium-scale systems,requiring few information exchange and has robustness against occasional communication failure.Compared with traditional dispatching methods,flexible resources such as energy storage,electrothermal conversion equipment,and thermal inertia of buildings can be fully utilized to cope with fluctuations in wind power output.(5)The power grid EOR software and integrated energy optimal dispatch software are developed.Oriented to the the provincial power grid,the power grid EOR software applies the EOR and multi-stage robust optimization theory to realize the three functions of renewable energy uncertainty analysis,power grid EOR calculation and dispatch plan generation.The displayed information includes the prediction error of renewable energy,the active power operation boundary of the system and the units,and the optimal generation plan.This software has been applied in Jiangsu and Ningxia operational centers.The integrated energy optimal dispatch software focuses on the regional integrated energy scheduling scenarios of industrial users.It applies the integrated energy robust optimization scheduling method to realize integrated cooling,heating and power modeling,integrated energy optimization scheduling,demand-side response and other functions,which can improve energy efficiency,reducing energy costs,and enhancing the ability to absorb renewable energy.This software has been deployed in an industrial park in Hangzhou.