Research On Decentralized Optimal Dispatch Of Power Systems Based On Value-Function Approximation

With the continuous development of power-related technologies,modern power systems are facing profound changes.For large-scale power systems,the interconnection of regional power grids has become increasingly closer.How to achieve decentralized optimal dispatch across the entire network in the premise of guaranteeing the independent operation of each area has yet to be resolved.In addition,as a clean renewable energy source,the penetration of wind power in large-scale power systems is rising.It is also worth exploring how to further consider the impact of wind power uncertainty on the decentralized optimization.As for the smart grid,with the development of communications automation and other technologies,it's important that how to make full use of the intelligent characteristics of each component and make it possible for each component to make decisions independently without losing the overall optimality.This paper studies the above three topics based on the value-function approximation theory.First,for the economic dispatch of multi-area power system,the spatial decomposition is realized based on the value-function approximation,and a novel decentralized optimization algorithm is proposed.The power transmitted by tie-lines is taken as the state variable which represents the connection between the areas.Then the corresponding value function is constructed to characterize the impact of each area's decision on other areas,so that the centralized model can be transformed into a decentralized model.Moreover,according to the convexity of the sub-problems in each area,the value function is approximated by the piecewise-linear function.Then the slopes of these approximate value function are constantly updated in the iterative solution process.The proposed algorithm does not require parameter tuning and has good convergence performance.Numerical simulations on several test systems and a real power system demonstrate that the proposed algorithm is favorable in terms of accuracy and adaptability.Then,considering the influence of wind power integration on multi-area economic dispatch,a decentralized stochastic optimization algorithm is proposed in this paper.The spatial decomposition is achieved by the analytical target cascading method,and the centralized optimization model of dynamic economic dispatch with wind farms is transformed into the decentralized stochastic sub-problems.The temporal decomposition is achieved by the value-function approximation theory,and the stochastic sub-problem in each area is decomposed into a series of single-time problems to be solved.The phase information of boundary nodes among the areas is used to coordinate their optimization schemes iteratively.Case studies on a 2-area and 6-bus system,the IEEE multi-area systems as well as an actual power system demonstrate the effectiveness of the proposed method.Finally,this paper proposes a new spatio-temporal decomposition algorithm to solve the dynamic economic dispatch problem in smart girds.The dynamic economic dispatch is transformed into a multi-stage optimization problem,and the spatio-temporal decomposition is realized by using the value-function approximation theory.In order to solve the problem of time delay caused by the existing update strategy of value function,an improved updating strategy is proposed in this paper,which can speed up the algorithm greatly.The proposed algorithm does not require parameter tuning and has good adaptability to information loss,plug and play,etc.The effectiveness of the proposed spatio-temporal decomposition algorithm is verified by the numerical simulations of several cases.