The Design Of Turntable And Recognition Technology Of Pointer Meters Of Inspection Robot

With the enlargement of network and the increase of substation level,intelligent inspection robot will play a more and more important role in substation inspection tasks.The design of turntable and implementation of meters recognition algorithm are two important parts of the inspection robot system.The turntable is the carrier of visual system.The recognition algorithm of meters is the key guarantee of normal operation of inspection robot.Firstly,the composition of system and major functions were recommended.The development situation of inspection robot at home and abroad was investigated.The key technology of inspection robot was summarized.Secondly,the mechanical structure design of different typical turntables was studied.The operating principle of turntables was analyzed.The advantage and disadvantage of different turntables were compared.Ultimately,an external rotor mechanism was used in the 2D turntable.The design of components(including horizontal shaft,vertical shaft,rotary structure,U-type frame and so on)was accomplished.The turntable's precision was analyzed and the error of different components was assigned.The natural frequencies and the deflection angle of the camera optical axis infected by vibration property of driving were acquired through the method of finite element analysis.The first six natural frequencies were distributed between 40.4Hz and 64.9Hz,which were beyond the external excitation frequency.The optical axis deflection angle of the camera was less than 0.223°.The designed turntable met the requirement of technical index.Thirdly,the meters recognition algorithm was implied.The typical objection detecting algorithms were introduced.Binarize image,acquire the image skeleton,detect the characteristic of pointer meters and recognize the reading with image processing techniques.The algorithms were optimized in the process of algorithm design according to the comparison result of different algorithms' advantage and disadvantage.The algorithm was simulated and implied with Open CV 2.4.10 based on Microsoft Visual 2012.Lastly,the validity and robustness of meters recognition algorithm were tested in experiment,the algorithm could adjust to the influence caused by variation of distance and light intensity.The absolute error controlled within 0.02 Mpa met the requirement of accuracy.Comparing the procession speed distinction of the proposed algorithm and other algorithms by experiment,the results showed that the method used in this dissertation could save 34.27% of the running time.The instantaneity of algorithm was promoted greatly so that the reading of meters can be acquired more quickly.