A Wind Power/Photovoltaic Typical Scenario Set Generation Algorithm And Its Application

With the higher penetration of renewable generation resources such as wind power and solar power, power system operations face new challenges as the supply uncertainty increases dramatically. Scenario analysis is widely used because it explicitly incorporates a probability feature of uncertainty. It is a practical limitation that a large number of pre-sampling discrete scenarios will reduce the efficiency of stochastic optimization model. In this paper, a generation algorithm of typical scenario set was developed to solve this problem. Firstly, the continuous probability density functions of wind power or photovoltaic generation were optimally transformed to discrete quantiles with precise probabilistic information based on Wasserstein distance metric. Then, with computational scales and loss of probability considered, the scheduling interval was reasonably divided into several fragments. All the scenario fragments were constructed to build an integral typical scenario set. This approach can reserve probabilistic information perfectly.To improve the operational performance of the algorithm, a revised K-medoids cluster technique was developed to realize the scenario reduction. From the point of view of cluster, the characteristic of scenario reduction is in low-dimensional, and it’s not sensitive to the amount of cluster. For the reduction of the calculation time, a parallel calculation schema named "CUDA" is imported. So the algorithm can solve large scale problem of scenario reduction in finite time and limit resource.Finally, a typical scenario set covering the entire scheduling interval was generated by iterative scenario reduction and fragment mergence operations. In the case studies, the precision and practicality of the typical scenario set are illustrated by using the real data of a distributed renewable energy generation demonstration project of the Guodian Yunnan Electric Power Co.