Research On Shaftkind Straightening System Based On Artificial Neural Network

Shaft-kind parts and components, which play important parts in mechanical devices , are inevitably bended and deformed during manufacturing . So it is indispensable that the shafts are minutely straightened in order to ensure their qualities.Based on previous researches and papers, especially according to the metal elastic-plastic mechanics theory ,this dissertation builds up the shaftkind parts' straightening model, analyses the relation among bending moment, curvature and deflection in the straightening process, and founds a suit of mathematics formulas of precise straightening .Take the YH40-25 System for a example, which is produced by HeiFei University of Technology, it is pointed out that the traditional Straightening Systems like YH40-25 have some merits but also have plenty of disadvantages by analyzing the system 's theory and structure. Based on this instance, a new straightening system must be set up, which can overcome the former one's demerits, in some extent.In the dissertation, a system including ANN module is established to guide Shaftkind parts' straightening, especially complex shafts, such as crankshaft, etc. Through inputting and learning a great deal of samples, the system performance reaches the initialized error's goal. And Simulation results indicate the new system can create proper straightening strategy and guide straightening technics successfully.