Research On Quick Inspection Method Of Operational Slope Based On UAV Photography

In order to solve the problems of high risk and high labor intensity in slope inspection during operation period in China,this paper relies on the maintenance project of Chongqing and Guizhou section of G75 Lanhai Expressway,and carries out a research on a quick inspection method of slope operation based on UAV photography.Through the analysis and research of the influencing factors of UAV photography,combined with the overall inspection and secondary inspection methods,the UAV inspection scheme is determined,and the slope model test is used to analyze the feasibility of the slope inspection scheme.Based on the inspection results,a set of slope disease image recognition and management system is developed,the benefit difference between manual inspection and UAV inspection is compared,and the feasibility of rapid inspection method for improving the efficiency of slope inspection is discussed.The main conclusions are as follows:1.Combined with the actual situation of the operation slope,the UAV inspection scheme is studied according to the analysis and judgment of the influencing factors of UAV photography,which mainly includes the research on the control scheme of photographic quality,the setting of photographic parameters,and the control scheme of flight quality.Photography parameters are ISO(speed)100,aperture f/11,shutter time1/125 s as the basis,under the condition of light change to adjust the aperture and shutter time,flight parameters selected course overlap 70%,side overlap 80%,flight height 100 m,to ensure the modeling effect of the slope.2.The main content and operation process of overall inspection and secondary inspection of UAV are studied.Combined with the inspection requirements,it is proposed that the inspection methods of UAV are mainly integrated inspection and secondary inspection,and the main content and operation process are studied.At the same time,the optimization analysis of UAV slope image processing method is carried out,and the slope modeling test is carried out by Mateshape software,and the feasibility of the scheme is verified.3.According to the requirements of operational slope inspection and maintenance,based on the concept of deep learning,the recognition method of deep learning neural network for slope disease images was studied.Matlab was used to construct,train and test the neural network,and the training and recognition accuracy of the neural network reached 83%.Different kinds of disease training accuracy are obtained,and suggestions for improving the model are put forward.4.The storage and management system of slope information was built based on Access database and VS2019 software.By creating database and establishing management system,the requirements of slope information query,management and inspection report export are achieved,which effectively improves the management efficiency of slope information and reduces the management cost.5.Uav inspection test was carried out on three slopes in Chongqing and Guizhou section of G75 Lanhai Expressway,and the traditional manual inspection and UAV quick inspection methods were compared and analyzed.The test results show that the UAV rapid inspection method can effectively improve the efficiency of slope inspection,and has good economic benefits.