Research On Robust Dispatch Of Wind Power Distribution Network Driven By Data Mode

With the severe challenge of global environment and its climate change,most countries and regions in the world are focusing on the sustainable development and efficient use of renewable energy in the energy sector.In order to solve the contradiction between renewable energy consumption and power system stability,a new type of active distribution network is established by adding distributed wind power generation to the distribution network while centralized wind power generation is connected to the grid.In this paper,we study the economic operation of the wind power distribution grid by using model-driven method to mathematically model the tidal network of the distribution network and data-driven method to predict the wind power and uncertainty,and establish the robust scheduling study of the distribution network of wind power jointly driven by data-model to reduce the economic operation cost of the distribution network while ensuring the voltage distribution and tidal distribution of the distribution network.The main research contents and the results obtained are summarized as follows:(1)For the radiation distribution characteristics of the distribution network,and the need to adapt to the access of devices including wind power and other adaptations to the development of new power systems,a distribution network model based on second-order cone relaxation is established and pairwise analysis is performed.Combined with the characteristics of the distribution network,the branch power flow model is used to model the distribution network.A distribution network model is derived based on second-order cone relaxation and its pairwise model using the second-order cone and its pairwise property.Simulation validation based on IEEE 33 standard nodes shows that the distribution network model based on the branch tide model has lower error and good adaptability than the distribution network model based on the Newton-Raphson method of iterating the conventional tide,while the original model has strong duality with the pairwise model,providing a model-driven approach for robust optimization of distribution networks with uncertainties.(2)In order to solve the problem that offshore wind turbine single-unit forecasting cannot quickly predict the overall power of offshore wind farms,and that uneven fluctuations among units can cause poor quality of cluster power curves and low forecasting accuracy,an ultra-short-term power forecasting model considering the spatial and temporal characteristics of offshore wind turbines is proposed.This method uses an improved dynamic time bending algorithm to quantify the similarity of spatio-temporal characteristics of offshore wind turbines and analyze the spatio-temporal characteristics of offshore wind turbines.Deep learning is used to predict offshore wind power.Offshore wind turbines are clustered and ultra-short-term power prediction is performed by considering the similarity of spatial and temporal characteristics of offshore wind turbines and bus location information.The analysis of the measured data of offshore wind turbines shows that the proposed method can effectively quantify and measure the spatial and temporal characteristics of offshore wind turbines and predict the power of offshore wind turbine clusters in the short and medium term.(3)In order to solve the uncertainty problem of renewable energy generator sets,especially the economic scheduling problem caused by the large number of centralized wind power and distributed wind power connected to the distribution network,a data model joint-driven robust scheduling of wind power distribution network is proposed.Modeling of distribution networks by branch flow model with second-order cone relaxation using a model-driven approach.A data-driven approach is utilized to achieve wind power output capacity prediction and error distribution for access to the distribution network,wind power uncertainty analysis,and load uncertainty analysis using Monte Carlo simulation.A joint drive approach using data-model is utilized to achieve minimized distribution grid costs considering distribution grid load,centralized wind power,and distributed wind power uncertainty.The simulation results show that three uncertainty scenarios are established using IEEE 33 nodes connected to wind power,which can effectively analyze the distribution network operation cost,wind power uncertain parameter distribution and the output power of wind power and the main network,while the robust sensitivity analysis conducted is beneficial to decision makers for flexible regulation based on their own sub-risk tolerance level.