Wind Field Reconstruction Based On Computational Fluid Dynamics And Data Optimization

Wind power,as a kind of renewable energy with low price and little impact on the environment,plays an increasing role in the energy system and has a great development potential.According to the prediction,wind power will be the largest single power source in the world by the middle of the 21 st century,instead of fossil energy.Although the installed capacity is increasing year by year,in terms of current forms of wind power generation,the share in the global energy structure is still low,and wind power generation still faces some problems.Wind prediction of wind power plants plays a key role in improving the efficiency of wind power grid connection.The traditional wind field prediction method based on historical data and probability statistics is difficult to obtain the accurate wind speed distribution field due to the randomness of wind speed prediction.While the calculation time of wind field prediction method based on physical model is too long,it is hard to reach the goal of real-time prediction.In this paper,a wind field reconstruction method based on proper orthogonal decomposition algorithm of hydrodynamics fusion features is proposed.The method adopted the thought of " trading space for time",because it takes time to air flow across the region.If the upstream wind speed is known,the accurate wind speed in the downstream can be known through calculation in advance,that is,the rapid reconstruction of wind speed field in the region has been completed before the wind reaches the designated area,achieving the purpose of real-time accurate prediction of wind speed,and providing a new way of thinking for future wind prediction.The main research contents and innovations of this paper are as follows:(1)A very small amount of data source information is used to complete the real-time and accurate reconstruction of the maximum wind field,and the method is verified both in simulation and experiment.First,multiple sets of computational fluid dynamics simulation of wind field based on terrain was carried out to obtain multiple sets of calculation results to form the original database.Then,the orthogonal feature decomposition algorithm was applied to the dimensionality reduction process of the original database.Then,the inverse problem principle was applied to realize the large-dimension reconstruction process of a small amount of data source information.In addition,the effects of truncation order,number of sensors and noise signal on wind field reconstruction are analyzed.(2)Two algorithms are proposed to optimize the location of data sources in the wind field reconstruction process.In order to solve the reconstruction process that will appear in the process of the calculation of the inverse problem of pathological and insufficient dimension into local optimal problems,as well as the reality considering the economics of needed by the reconstruction process,put forward two different data sources under the condition of choosing optimum algorithm:1)in the laboratory under ideal conditions,can be process as the matrix of the inverse problem of underdetermined equations to solve the problem of condition number will affect the orthogonality of the matrix and stability,can put forward by looking for the constraint matrix every step in the calculation of minimum condition number is finding ways to optimize the location of the data source;2)In the actual environment where the error exists,the influencing factors of the reconstruction error are found through mathematical derivation,and a method is proposed to find the maximum eigenvalue of the constraint matrix in each step calculation so as to find the optimal location of the data source.In the simulation and experiments for two kinds of optimum data source method and random selection of data source for the effect of the wind field reconstruction were compared,the results show that when using the same number of sensors and the same set of snapshot matrix,two kinds of optimization algorithm to calculate the measuring point position for the optimization of the wind field data reconstruction results are better than that of random point data of wind field reconstruction results,and the universality of the second optimization algorithm of the is higher.(3)An efficient and rapid real-time reconstruction method is proposed based on the basic wind field reconstruction method,aiming at the problems such as large computational domain and long time in the process of using computational fluid dynamics principle to simulate the complex terrain.This rapid reconstruction algorithm can not only greatly save the amount of data and computing time in the real-time reconstruction process,but also save part of the off-line work.Compared with the wind field data of computational fluid dynamics simulation or reconstruction directly using rough grid,the wind field reconstructed by this method has better effect and higher resolution.(4)A complete set of wind tunnel test platform was designed and built,and all activities including wind tunnel design,fan and sensor selection,data acquisition programming and experimental design were completed.In order to solve the problem of uniform flow,a guide plate and a steady flow device are specially designed in the diffuser section to ensure that the wind tunnel has sufficient space to complete a variety of sensor arrangements and wind speed information acquisition at different heights.A series of experiments are designed to verify the two wind field reconstruction methods and two optimized data source selection methods proposed above.Through the comparison and analysis of the relative errors between the real-time measurement of wind speed and the reconstruction of wind speed at the verification point,the feasibility of the proposed method in reality is proved.