Research On Machine Vision Monitoring Techniques For Running State Of Wind Turbine Blade

Due to the complicated operating environment,the wind turbine blade,a kind of large and slender component with flexibility,may emerge varying degrees of excessive deformation even breakage under the long-term action of multiple running loads,which poses a serious threat to the safe operation of the wind turbine.Therefore,it is of great significance to detect structural damage and diagnose the operating conditions of wind turbine blades.However,the current monitoring techniques for running state of wind turbine blade have deficiencies,such as limited by the actual application scenario,lacking of adequate research on failure mechanisms and insufficient failure criteria,and cannot be widely applied in practical projects.Combined the actual requirements of operation state monitoring and fault diagnosis with machine vision technology,which has the advantages of large scenes,long distances,high flexibility,this article analyzed the structure of wind turbine blade and deformation caused by the loads during operation,then conducted a research on machine vision monitoring techniques for running state of wind turbine blade,and finally proposed two kinds of condition monitoring method——the monitoring method based on the mark deviation of inter-frame,and the monitoring method for running state of wind turbine blade based on local pose estimation.The monitoring method based on the mark deviation of inter-frame sets marks at the tip of each blade where the blade deformation is most prominent,and acquires the images of the two blades to be tested sequentially rotate to the image acquisition position by using the monocular camera respectively.After processing the images of the blades,we can calculate the position and geometrical deviation of the tip marks between the two-frame images(representing two adjacent blades),which is caused by the geometric deformation between the adjacent blades.By comparing the deviation with the threshold,the level of the blades relative deformation can be obtained.Furthermore,in order to improve the accuracy of the monitoring method,the threshold is corrected according to the stability of the operating state of the impeller.Another monitoring method is the monitoring method for running state of wind turbine blade based on local pose estimation.This method adopts the same tip mark setting and image capturing position selection as the monitoring method based on the mark deviation of inter-frame.Firstly,a wind turbine coordinate system for describing the tip marker and the camera's three-dimensional pose information is established,and the relative pose transformation matrix between the blade marker region and the camera in the wind turbine coordinate system is derived(containing unknown parameters indicating the degree of deformation of the blade).Then,the monocular camera is used to obtain the tip mark image of the blade to be tested rotate through the image capturing position,and the relative pose information and the transformation matrix between the tip mark area and the camera are obtained by the pose estimation method.Furthermore,the parameter indicating the degree of blade deformation can be obtained through the correspondence between the transformation matrix of the tip mark area and the camera in the wind turbine coordinate system and the transformation matrices obtained by the pose estimation method,realizing the monitoring of the running state of the wind turbine blades in service.An experimental model is built to simulate different operating states of the wind turbine to analyze and demonstrate the effectiveness and reliability of the proposed monitoring method.Using the image processing programs analyzes the blades with different degrees of deformation.The experimental results show the effectiveness and reliability of the above method.