Study On Some Key Issues In Intelligent Control Of CNC Machines

Recently, Numerical Control Technlogy has been playing a more and more important role in industrial fields such as CNC machine, robot, electrical discharge machining, precision electronic manufacturing and so on. In Nuerical Control research area, some key problems including high performance motion control technology, hierarchical intelligence, complex trajectory catch a close attention among academia and industrial. From the viewpoint of open intelligent CNC platform, this thesis mainly discusses realtime operation system construction, NURBS approximation with biarc, Look-ahead technique, Genetic algorithm improved pid controller and BP neural network interpolation. The related research work both in theory and practice is performed with expectation to make progress in intelligent control application in CNC fileld. The proposed idea is briefly listed as below:Considering the characteristics of current electromechanical devices, a new method building high performance industrial motion control platform is presented in detail. Intergrated with RCS theory, some software module defined as TASK, MOTION, GUI, IO are designed for development purpose. The research work was carried out in both software and hardware. The realtime Linux kernel port technology is investigated and the process scheduling time can be limited to microsecond. On the other hand, motion control IC and some function simulation experiments are investigated to fulfill development purpose. Finally, the new type of hardware platform with eight axes function control (up to sixteen axes) applying to both open and close loop control system was implemented to verify the research feasibility.Since the complex trajectory (Non Uniform Rational B-spline) interpolation technology has become a general standard in CAD research filed, more and more development efforts were put into this area. This paper presents a new method building NURBS interpolation technology platform from the viewpoint of theoretical deduction of NURBS mathematic relation and numerical algorithm of biarc. And some numerical methods to reduce speed variation are investigated to fulfill development purpose. Finally, the experiments are carried out to verify the research f easibility.Also the trajectory control and optimization are proposed to deal with both lineal and circular micro-segments. At the same time, Look-ahead technology proposed in this thesis is employed to predict the speed variation of corner crossed by two adjacent segments. Consequently, an adaptive federate adjustment can be attained which will enhance the machining efficiency by accelerating or deceleratting the desired tolerance. Based on the above, a novel S-shape Acc/Dec algorithm equipped with rounding error compensation tactic is employed which compensates the error at the transition of each segment. Moreover the proposed algorithm can reduce quantization error effectively.In order to investigate the intelligent numerical algorithm application in motion control engineering work, this thesis mainly discusses the research and implementation work of BP neural network interpolation and GA improved PID controller. In neural network training study, BP technology was employed to approximate the NURBS curve and transform problem from parameter space to geometrical space. Moreover, simulation platform was built to gain high performance of GA improved PID controller, and comprehensive experiment was carried out to verify the proposed algorithm.