Research On Bladeautomated Assembly And Image Recognition

With the rapid development of manufacturing industry, the research and application of automated assembly technology is more and more widely. This issue is also generated in this context. It comes from the need of arbor automated assembly in a shredder plant. The arbor automated assembly and blade image recognition are the main contents of this paper. This thesis first describes the subject in detail, and then gives the key issues which are need to be solved. These issues include the feeding problems, identifying the front and back side of blade and blade alignment and running-in.In the designing process of automated assembly system, the author introduces the general requirements of overall layout design and factors to be considered. And on this basis, the paper compares and analyses the proposals put forward in the designing process. At last it chooses the design which uses guide bar and moving feed. This design is easily layout and is economical. Then the author designs the automated assembly module of blade in detail. This module contains blade displacing mechanism, blade alignment mechanism and blade running-in mechanism. Based on the analysis of functional, the author determines the design proposal and gives the structure plans. Movement simulations are also given.Blade image recognition is the core of this paper. Based on neural network, the author transforms the RGB picture to gray and filters it. Then extract the binary image and the key parts of a cutter by using matlab software and image processing knowledge. Pick up the moment invariants of the cutter image as the input of neural network. It also selects the type of BP network at the basis of analyzing the type of artificial neural network and gives the BP algorithm and its improvement. Based on this, the author constructs neural network system to recognize the blade. The experimental results show that the system can identify a blade very well.