Technology Of Abrasive Water Jet Precise Machining Based On Open Numerical Control System

Abrasive water jet (AWJ) presice machining technology is studied detailedly. A neural network model of abrasive water jet presice machining is established based on artificial neural network (ANN). The visualization interface for computation of neural network is developed. As a calculation unit, the COM component of MATLAB is integrated into the software. Therefor, computation and data analysis as well as simulation of abrasive water jet precise machining can be realized. It simplifies programming the algorithm of artificial neural network. Also, it provides the foundation for integration of ANN model of AWJ with computer numerical control (CNC) system.The open CNC system with double CPU based on PC and PMAC (Programmable Multi-Axis Controller) module is put forward. The system has "PC+PMAC" module. The numerical control system for abrasive water jet machining is developed using Visual Basic and COM technology.The following work is included in the thesis:1. An ANN model for abrasive water jet based on artificial neural network theory was established. The feed speed for the abrasive water jet presice machining can be predicted precisely through this model.2. A module for computation of ANN algorithm based on database, COM component of MATLAB and Visual Basic was developed. M function for training neural network is a component that is generated by COM Builder of MATLAB. Thereafter it can be called as an upper server by Visual Basic.Seamless integration of Visual Basic and MATALB are realized through COM component and M function. Consequently the calculation of artificial neural network is complished.3. The sampling interpolation is employed and studied in the open numerical control system for abrasive water jet machining. An open NC command system based on PMAC is studied. System parameters setting and programming for PMAC are carried out.4. Based on the advanced software developing method, the primary COM-CNC software platform is designed using object-oriented analysis and behavior-based model.