Research On The Key Technology Of Automatic Feeding Device For A Kind Of Ceramic Part

Snap-action thermostat is the key part for temperature control and overheat protection used in small household electrical appliance,it has a stable performance and a long service life.In the front assembly line of snap-action thermostat,the ceramic shell is fed by vibrating feeder now.But this feeding method has such problems:the friction between metal and ceramic leads to the shell being black,the dust caused by the vibration leads to the product's poor conduction,the shell is easy to stuck and the operating noise.These problems have become the bottleneck of improving the product quality,so it's urgent to develop a new kind of feeding machine which is vibration-free,shell-not-being-black and stuck-free.This paper designed a kind of automatic feeding system which can meet the above requirements,the system includes three modules:feeder,image recognition system and part delivery manipulator.The feeder separates and initially orients the ceramic shell,the image recognition system recognizes the six gestures that are initially oriented and sends out instructions to the manipulator which finish orienting the ceramic shell and transmits it.The key technologies of the modules have been explained.For the feeder,it has been analyzed completely from structure design,simulation to experiment verification.The cam mechanism and reclaiming performance is analyzed by simulation based on ADAMS.The reclaiming model for simulation is built using secondary development technology of ADAMS.Then the prototype is built for reclaiming experiment,by the statistical analysis of the experiment result,the components' operation process of the prototype is improved and the reclaiming success rate is improved further.For the image recognition system,the image recognition program is developed.The image is processed by graying,median filtering,thresholding method,then the Euler number of the binary image is computed which can divide the image gesture into upright and side gestures.The edge is extracted for the upright gestures,then the edge image is processed by cropping and Hough transforming,after that the eigenvalue is extracted.The threshold is obtained by analysis of many samples.By comparing the eigenvalue and the threshold,the two upright gestures are recognized.For the side gestures,four eigenvalues are obtained by scanning the image from horizontal and vertical directions,by finding out the biggest eigenvalues,the four side gestures are recognized.The image recognition program is realized by MATLAB in the earlier period and is wrote by C language using Visual Studio in the later period,seventy-two test pictures are recognized by the program and no recognition error is found.For the part delivery manipulator,the Festo grasping system is selected which can move in the XY direction,and the wrist of two degrees of freedom is designed,it can adjust the reclaiming posture according to the six different gestures of the ceramic shell and then rotate the ceramic shell to the same required gesture.