| High penetration of the new energy resources,inter-connected structure between regional power grids and the improvement of the dispatchability in load side have become three key characteristics of modern power grids.It brings not only more potential,but also higher challenges to the ability and technology of stable and efficient operation of power grids.As an important technology to support the power grid system,power dispatch needs to gradually adapt to and utilize these characteristics.However,there are relatively insufficient researches on the dispatch optimzation of the interconnected power grids considering simultaneously the uncertainties and dispatchability of the source and load side.Therefore,based on the rapid development of new energy resources and flexible loads,this thesis focuses on the day-ahead unit commitment and the intra-day economic dispatch cases,and studies the dispatch optimization of the inter-regional power grid under the uncertain environment considering the coordination of source-grid-load side.Due to the focused problem being uncertain,multi-dimensional,non-linear,etc.,a hierarchical mechanism is adopted in this thesis.Then,the corresponding dispatch optimization methods are studied according to the characteristics of different problems.Specifically,works in this thesis include:(1)Day-ahead dispatch optimization of inter-regional power grid based on improved discrete-continuous particle swarm algorithm considering the participation of DLC flexible load.In this work,the coordination dispatch optimization of inter-regional power grid with the participation of DLC flexible loads in day-ahead time scale is studied.First,based on the hierarchical dispatch mechanism of the inter-regional power grid in China,a coordinated dispatch mode of the source-grid-load side for the inter-regional power grid is constructed that takes into account the tie-lines power transmission.Then,concerning the uncertain factors affect the constraints and the values of objective functions,the stochastic chance-constrained programming method is used to formulate the basic dispatch optimization model.At last,considering the difficulty of finding the optimal dispatch plan in the basic model,an improved hierarchical optimization model which consists of an upper layer for the unit commitment and tie-line schedule,and a lower layer for the economic dispatch is established.And,two types of improved particle swarm algorithms are used to solve the upper and lower models respectively.In the example analysis,two IEEE 39 bus systems and two IEEE 118 bus systems are used to verify the proposed model and methods.The economy,cleanliness and security of the obtained dispatch plans are evaluated and analyzed as well.(2)Hierarchical learning optimization study for the intra-day coordination dispatch of interconnected power grids considering the participation of two types of flexible loads.With the continuous improvement of the power market,the role of power dispatchers gradually changed.The service attributes of power dispatch are recognized more and more widely.Therefore,focusing on the concept of the quality of service(Qo S),the hierarchical-learning based optimization of the intra-day coordination dispatch for the inter-regional power grid system with two types of flexible loads.First,the response model of the price-sensitive flexible load is described.Involving the response behavior and satisfaction degree of the price-sensitive flexible,a dispatch service quality index and its evaluation model is proposed.The multi-objective coordination dispatch model of the inter-regional power grid considering Qo S is formulated.Then,a model-free hierarchical optimization method based on the learning technique is designed.The hierarchical structure consists of two levels and multiple agents,where the tie-line dispatcher and the regional power grid dispatcher are viewed as the upper agent and the lower agents separately.An improved reinforcement learning algorithm is adopted to find the optimal dispatch policy for each agent.At last,the simulation results are shown to validate the effectiveness of the designed method.Specifically,the optimization process and the obtained dispatch policies are analyzed,and the impact of the Qo S index on the optimization is introduced.The trends of multiple agents in the learning and optimization processes,and the impact of Qo S index on the system optimization are analyzed.At the same time,the optimization results for the case based on two 118 bus systems in Chapter 3 are given and analyzed.The rationality of the obtained intraday dispatch plan is validated as well.(3)Deep reinforcement learning-based optimization hierarchical method of the intraday power dispatch for inter-regional power grid considering the participation of load-side operator.Considering the continuous development of load-side operators,the coordination dispatch optimization method based on deep reinforcement learning algorithm considering the autonomous decision-making microgrid is studied in this work.First,the autonomous response model of operator is formulated by solving its optimal energy management strategy.According to the dispatch process and mode of inter-regional power grids,the coordination dispatch model of multi-regional power grid with the participation of load-side operator is formulated.Then,an improved deep reinforcement learning algorithm based on a hybrid exploration mechanism and a double deep Q network is designed.Based on the learning-oriented hierarchical structure designed previously,a deep reinforcement learning-based dispatch optimization method of multi-regional power grid is proposed.At last,the effectiveness of the proposed method is validated by the IEEE 300-bus system.Furthermore,the results obtained by the designed hierarchical method,conventional centralized deep reinforcement learning method and the hierarchical reinforcement learning method are compared.The comparison results help to show the superiority of the proposed method.In addition,for the unified case based on two IEEE 118 bus systems in the first two chapters,the corresponding learning results optimization are given,and the rationality and effectiveness of the obtained intra-day dispatch plan with correction are analyzed. |
