Machine Vision Based Automatic Assembly Technique For Frequency Stabilization Device

The assembly of microminiature precision devices for aviation is not highly automated at present,the expected precision is achieved mainly by manual assembly.Although manual assembly can flexibly deal with various complex assembly situations,the problems remain such as poor consistency of product performance and low production efficiency.With the service environment becomes more and more complex,it makes greater demands of assembly accuracy.In addition to the lack of highly skilled operators and labor cost increase also limits the improvement of the capacity,so it is necessary to research automatic precision assembly technology,use machine vision for automatically measuring and compensating,and finally realize the microminiature precision assembly.The automatic alignment assembly of thin section parts is a common problem in the automatic assembly of aviation precision devices.Such components exist in accelerometers,gyroscopes and other inertial navigation devices.In order to realize automatic precision assembly,this paper develop the automatic assembly system for frequency stabilization device based on machine vision,which realized the automatic assembly of slice components and metal brackets.Firstly,we analyzed the assembly task and requirements,and designed the assembly process of components.Based on the modular idea,the automatic assembly system was divided into visual measurement module,assembly operation module,feeding platform and base module and computer human-computer interaction module.Customized assembly component structure,selected the hardware such as controller,sensor and driver,and built the precision automatic assembly system.Using MFC framework to write hardware debugging software to realize the connection and control of the hardware in the system.According to the requirements,based on Windows NT,the use of C++ language and the OpenCV library to write software.The software is divided into interactive layer,task layer,logic layer and behavior of four layers architecture,integration of visual inspection,image processing,data acquisition,serial communication and other functions,automatic parts size and position detection,and finish the assembly tasks.The key of automatic precision assembly is to obtain the position information of each part and calculate the displacement of the part in place automatically.In this paper,a part contour edge extraction algorithm is developed to calculate the geometric information of the part edge.Some of the parts in this study are larger than the camera's field of view,so the image mosaic algorithm is programmed to calculate the information.The error compensation model was established and the camera pixels were calibrated to improve the calculation accuracy.Then,the Angle between the visual measurement module and the guide rail in the assembly operation module is calibrated,and the compensation algorithm is written to reduce the assembly error caused by the Angle deviation of the guide rail.Meanwhile,the Angle between the guide rail and the compensation algorithm are also introduced into the image mosaic to realize the measurement of the key feature dimensions.Finally,in order to reduce the random error caused by the adhesive parts when the adsorption head is raised,a hard PTFE adsorption head is designed and made to replace the butene-butene rubber adsorption head.The experimental results show that: the automatic assembly device can realize the transition matching assembly between the metal bracket and the base,and the assembly coaxiality of the thin component decreases from 292.4?m to 19.6?m.