Research on Data-Driven Fault Location for Inverter Based Resources Interfaced Power Systems

Speaker:  Zhiqiang Duan

Time:      14:00, Mar. 27

Location: SIST 1A-200

Host:      Yu Liu 

Abstract:

At present, the main goal of the new power system is to gradually replace fossil energy with renewable energy such as wind power and solar power. However, wind power, photovoltaic and other renewable energy systems are mainly based on power electronic control, and their system inertia is smaller than that of traditional power generation. At the same time, it also has strong volatility and randomness, and large-scale access to the power grid will bring great challenges to the stable operation of the power grid. Therefore, accurate prediction of new energy output is an important guarantee for intelligent dispatching and safe operation of power grid.

The continuous development of artificial intelligence technology provides more accurate and effective methods for the prediction of new energy output. Data-driven technology has higher accuracy and universality than traditional physical model driven and statistical model. This talk will start from different scenarios of wind power and photovoltaic, and use the method of combining physics and artificial intelligence to obtain accurate wind power and photovoltaic output prediction results.

For multiple wind farm stations in a large-scale wind power cluster, this talk builds a wind power physical model according to the physical process of wind power generation, uses data-driven method to extract time and space features, applies numerical days forecast data to ultra-short-term prediction, and combines the two to achieve ultra-short-term wind power prediction for multiple stations. For multiple photovoltaic sites in the cluster, a two-stage graph convolutional neural network is used to extract meteorological information features and location information features respectively, which is used to improve the accuracy of ultra-short-term photovoltaic output prediction. For the small-scale photovoltaic power generation system in the park, accurate physical modelling of the photovoltaic power generation system is carried out, and the ultra-short-term photovoltaic power output prediction in the park is achieved by combining with data-driven methods, and a simple algorithm landing attempt is carried out on the 2060 platform of the school.

Bio:

Mr. Zhiqiang Duan joined PSPAL in March 2022. He is currently a master student in PSPAL (starting from September 2022). He receives his B.S. degree of Electronic Engineering from ShanghaiTech University, in Summer 2022. His research interests include Generation forecasting of new energy systems.