Research On Optimal Operation Of Electrical Power System

It would create great economic benefits if the operation of a power system be optimized due to its large amount of input and output, so it has become one of the most important topics for electrical utilities. Among various problems in optimal operation of power system, power flow analysis is a fundamental one, making the optimization results more compatible with the operation of real power system; medium/long-term unit commitment (UC) is an expanded problem of classic short-term unit commitment, to realize a more economic UC result in a longer time horizon; optimal operation of energy storage system is a problem accompanied with the development of power system. Therefore, the work of this dissertation is based on the above-mentioned aspects. Details are as follows:Firstly, two methods are introduced to calculate the complete solution of power flow equations. One is the so-called resultant method, and the other one is eigenvalue/eigenvector method based on Grobner bases. In resultant method, for each unknown variable, a high-degree polynomial about this variable can be given, and the values that make the polynomial equal to zero are exactly the solution of this unknown variable. Eigenvalue/eigenvector method based on Grobner bases transforms the original problem to eigenvalue/eigenvector calculation of a matrix. Compared with elimination methods, both proposed methods can avoid cumulative errors, and remarkably, eigenvalue/eigenvector method can solve systems more effectively than elimination methods. It is worth noting that both methods show perfect methodology on solving polynomial equations and resultant method can give out analytical expressions among two variables. Therefore, the two method have potential value in many areas of power system research.Secondly, with regard to medium/long-term UC problem, considering utilization of units should be well organized and calculation time should be acceptable, a two-stage medium/long-term model is proposed. This model is derived from traditional UC model based on Benders decomposition, dealing with some key problems of decomposition with consideration of power system operation features. A medium term UC problem and a long term UC problem are studied by both proposed two-stage model and traditional model. It is concluded that:1) The optimization results of proposed model are good and acceptable;2) The efficiency of the proposed model are proved since it costs less time to solve medium term UC and obtains optimization results in reasonable time while the traditional model can not be able to. Thirdly, in the aspect of optimal operation of energy storage systems, both offline and online situations are taken into account.During offline optimal operation research of ESS, two scenarios are considered:1) a single battery energy storage system (BESS) used for peak load shifting and power loss reduction;2) several BESSs used for peak load shifting and minimizing spilled wind energy in a distribution network with high penetration of wind power. Both problems are solved by two-layer optimization. In outer layer, genetic algorithms are used to optimize BESS charge/discharge status, which are optimization problems on discrete variables; in inner-layer, quadratic programming (QP) are used to optimize BESS charge/discharge power. Using the variance of load in a given period as the objective function, in scenario1the QP in inner layer is always convex, while in scenario2the QP in inner layer cannot be guaranteed to be convex. Thus a small modification is made on the objective function to make it always a convex QP. The effectiveness of proposed methods is proved by the case study results.The distributed optimal control strategies of multi-ESS are studied in the time-domain operation. Assume that communication networks are used to connect all ESS together, so that all ESS can operate automatically and meet the response rate need of real time operation. Two communication network topologies with consideration of so called "N-1" rule are suggested for the situation of small and large amount of ESS respectively. Under the suggested topologies, distributed consensus algorithms based on equal incremental rate of power output are presented to revise the results of offline optimal operation. It is demonstrated that the proposed algorithms could realize the optimal control objectives effectively in a distribute way, which proves the rationality of the suggestions on communication network indirectly.