High Speed Wind Turbine Blades Deformation Measurement Based On Digital Image

Wind energy has attracted much attention in the field of new energy for its renewable,non-polluting and mature application technology.Moreover,the installed capacity of wind turbines in the world is developing in an upward trend.However,the wind turbine blades are subjected to the complex force are easily damaged and thus it causes some accidents.Therefore,in order to avoid the economic losses,it is necessary to measure the deformation of the wind turbine blades for accident prevention.Among the existing measurement methods,the ones based on machine vision have been widely used due to the non-contact,high precision,and convenient measurement system construction.So the 3-D digital image correlation method is used to measure the deformation of the wind turbine blades in this paper.Although the digital image correlation method can obtain the whole field strain,it still has the disadvantages of huge computational complexity and slow calculation speed.Therefore,based on the realization of the exact matching algorithm and the completion of the 3-D measurement method,this dissertatio n implements the algorithm based on the characteristics of the algorithm itself.Heterogeneous parallel acceleration operation based on GPU ensures the accuracy of deformation measurement and improves the system speed.The paper makes an analysis of the necessity of measuring the deformation of wind turbine blades and compares various measurement methods.3-D DIC method is selected.Based on the basic principle,the matching algorithm of the 3-D DIC method is studied from the aspects of the selection of correlation function,sub-pixel matching method and matching parameters,and the feasibility of the algorithm is verified.The paper investigates the measurement scheme of 3-D deformation.The 3-D deformation measurement based on stereo vision includes two parts,which are camera calibration process and 3-D reconstruction and deformation fitting process.Furthermore,according to the characteristics that each sub-region does not interfere with each other in the image matching process,the acceleration method of the matching process based on the GPU is studied,and the heterogeneous parallel computing program is designed.Finally,Experiments were performed using stereo vision system,measureme nt programs,two-dimensional motion mechanisms and total stations.The experimenta l results show that the proposed method can measure 3D deformation with high precision.Comparing the matching algorithm between serial and heterogeneous operation modes,it is proved that the matching algorithm using GPU parallel acceleration can obviously increase the matching speed.