| Transmission line icing is a natural phenomenon,electrical accidents such as conductor galloping and broken strands or lines of transmission conductor,icing flash-over of insulator,tower collapse and others often occurred under icing conditions,which directly affect the safe operation of power grid and social stability.Scholars at home and abroad have conducted a series of studies on the icing mechanism,icing monitoring technology and icing protection methods of transmission lines.Visual sensors have been widely used due to its simple installation and intuitive results.However,most image analysis methods based on vision sensors mainly focus on the measurement of the icing thickness on the conductor and insulator,but rarely involve the measurement of icing type,icing density,icing shape,and degree of icicle bridging,and other parameters that are closely related to the hazard of transmission line icing,these parameters are also very important for icing theory and ice melting technology in icing disaster scenario.Therefore,this dissertation takes the image processing technology as the research means to study the methods about how to measure parameters such as the icing type,icing density,icing shape and bridging degree of icicles on transmission lines,and provides a systematic solution from the front-end image acquisition to the back-end software analysis,which is of great significance to improve the safe operation level of transmission lines under icing conditions.The main contents of this dissertation contain the following aspects:1.In order to improve the quality of icing images collected in harsh climates,an icing image enhancement method based on adaptive low-frequency attenuation defogging is proposed to realize the enhancement processing of the foggy icing images with low definition,insufficient lighting,and low contrast.Based on imaging mechanism under atmospheric light and characteristics in frequency domain of image,a spatial adaptive attenuation factor is constructed and introduced to the low-frequency component of foggy icing image to complete the defogging processing,and the high-frequency component is sharpened to enhance the image details.It is pointed out that the reconstructed image based on the lowfrequency defogging component and the high-frequency sharping component can obtain better defogging enhancement effect,and its overall contrast can also be effectively improved.Compared with classical defog enhancement method,the test results show that the proposed method has robust robustness and wide applicability to multiple transmission lines icing scenes with different objects and different backgrounds.2.As to find the feature parameters that can characterize different icing types on transmission lines,a texture enhancement method based on local difference processing is firstly given to highlight the surface texture of icing images.From the histogram distribution characteristics of texture enhancement images of different icing types,eight feature parameters of statistical texture that can distinguish different icing types are selected.On this basis,an automatic recognition method of icing types based on histogram multi-features and LIBSVM is proposed,and the icing types recognition model based on LIBSVM is constructed and trained,and the effectiveness of the model is verified by a series of artificial icing experiments.The experimental results show that this method can effectively distinguish between three icing types of soft rime,hard rime and rain rime,and the average recognition accuracy is as high as 96.67%.3.Aiming at the irregular icing shape of transmission conductor under natural icing environment,a recognition method of conductor icing shape based on vision sensor and force sensor is proposed.Firstly,the diameter distribution in the front view and the top view of the ice-coated conductor is extracted and measured by the vision sensor and image analysis,and the average diameters in the two views are used to construct the minimum rectangular envelope of the cross-section shape on the conductor.Secondly,the equivalent area of the cross-section of the ice-coated conductor is calculated by using the force sensor and the built-in mechanics calculation model.Finally,under the common constraints of the minimum rectangular envelope and the certain area,an algorithm based on iterative curve fitting is given to describe the cross-section shape of icing conductor.In addition,shape factors such as roundness and overlap rate are defined to evaluate how close the fitted conductor icing shape is to the measured conductor icing shape.The experimental results show that this method can identify and describe the conductor icing shape at different stages in the icing process,as well as the conductor icing shape under different icing types.The overlap rate between the conductor icing shape fitted by algorithm and the shape measured by manual can exceed to 0.85.4.In order to evaluate the icicle bridging state of ice-covered insulator,a recognition and evaluation method of icicle bridging for suspension glass insulators by image analysis is proposed.Firstly,the icicles are effectively enhanced by the improved texture filtering operator,and then segmented by using the improved local standard deviation threshold.At the same time,the insulator sheds are separately enhanced and segmented by the normalized color difference operator and the maximum inter-class variance threshold.Secondly,a calculation method of icicle length is proposed based on the smallest axis-parallel enclosing rectangle,as well as a calculation method of two adjacent insulator sheds spacing is proposed based on the ellipse fitting.Finally,the evaluation rules of icicle bridging state are formulated by combining icicle length and insulator sheds spacing.The field test results demonstrate that the accuracy of the method for evaluating the icicle bridging condition of ice-covered insulator can reach 92.17%,which can meet the requirements of practical applications.Furthermore,in order to explore the feasibility of the above methods in the on-site icing condition monitoring of transmission lines,the processing effect and performance of the proposed algorithms for the real transmission line icing images are analyzed in this dissertation.The system was installed and tested on the 110kV outdoor transmission lines platform,and the recognition efficiency of each functional module integrated in the backend expert software is completed,including the icing image enhancement,icing type recognition,conductor icing shape recognition,and insulator icicle bridging condition recognition,which confirmed the effectiveness and feasibility of the technology.In this dissertation,the image recognition and condition evaluation methods of icing state quantities such as the icing image defogging,icing type,icing density,conductor icing shape and insulator icicle bridging of transmission lines are studied,which provide a new detection idea and technology for the accurate acquisition and reliable evaluation of the transmission lines icing condition,and also provide a new basis for the acquisition of key parameters in transmission lines icing theory and ice melting technology. |
PDF Full Text Request