Research And Development Of Intelligent Inspection Vision System For Substation Equipment

The daily inspection of the traction substation is very important to ensure the stable operation of the electric locomotive.At present,many substations in China use the method of manual inspection to patrol the relevant equipment of the substation.However,this kind of inspection method has problems.Inspection results are easily interfered with personal factors such as the worker's experience level and responsibility.Therefore,it is necessary to develop an intelligent identification system to complete the automatic identification of the traction substation based on the image processing technology.Then the testing results are delivered to the control room.Based on the intelligent identification of substation,according to the situation of the site,the scheme of the magnetic guidance AGV inspection vehicle equipped with intelligent identification system is proposed,the magnetic guided AGV trolley is used as the mobile platform,the industrial camera is used as the image acquisition device,and the industrial computer is used as the image processing medium.The process of intelligent inspection is analyzed,the structure of the inspection platform is designed,and the relevant hardware is selected.On this basis,the status of different indicators on the substation,the number of pointers and the number of digital display representations were studied.For the identification of the circular lights,firstly the light image is transformed into HSV space,and secondly,the Indicator light is extracted with different colors,and then for the circular lights of different colors,the Hough transform is used for circle recognition.For the deformed indicator,we change thresholds to handle it separately.Finally,we determine the indicator lights' status according to the V value near the center point.For the identification of the pointer meter,firstly,a series of processing is performed on the collected instrument image including Grayscale,Image smoothing,and an adaptive threshold method is proposed to complete the segmentation of the instrument image,and secondly,the tilted instrument image is corrected by the Hough transform method,the pointer region is extracted from the corrected image,the silhouette method and sewing method are mainly studied.Then,the extracted pointer is refined by the refinement algorithm.Finally,the fitting of the pointer line is completed by the Hough transform.According to the linear relationship between the angle and the instrument representation,we complete the reading of the pointer number,and analyze the errors that may occur during the reading process.For the reading of the Digital display,we mainly research two key issues in the reading of digital display: numeric character segmentation and numeric character recognition.According to the possible phenomenon of character sticking in the recognition process,a segmentation scheme based on projection method for fixed characters is proposed.On the basis of analyzing the advantages and disadvantages of these two identification methods,the improved sewing method is used to complete the identification of numeric character.According to the algorithm studied in the previous,the intelligent identification visual system software is built,and the structural scheme of the upper computer plus the lower computer is established.The upper computer part includes the design of the human-machine interface,the design of the data structure,the design of the communication system,etc.The lower computer part includes the design of control program,and it can realize the motion control of the motor.Finally,a visual system is developed to complete the intelligent inspection of the substation equipment.