Intelligent Control for a Robot Belt Grinding System

A robot belt grinding system provides promising prospects for relieving hand grinders from their noisy work environment, as well as for improving machining accuracy and product consistency. However, for a manufacturing system with a flexible grinder, controlling the robot to perform precise material removal from free-form surfaces is a challenge. In the belt grinding process, material removal is related to a variety of factors, such as workpiece shape, contact force, and robot velocity. Some factors of the grinding process, such as belt wear, are time-variant. To achieve the desired removal in the grinding process, an intelligent control method for the industrial robot is proposed in this paper. First, an adaptive grinding process model that can track discontinuous changes in working conditions is constructed to precisely predict material removal in accordance with in situ measurement data. With incorporated prior knowledge, the method considerably improves model accuracy, which worsens when new samples from an in situ measurement are insufficient or are unevenly distributed under new working conditions. After this, an online trajectory generation method for the robot control parameters is proposed. By calculating the optimal control parameters in real time, the control transition process is shortened and its negative effect on grinding quality is reduced. Finally, the preliminary grinding experiments validate the workability and effectiveness of the proposed control method.

[1]  Corinna Cortes,et al.  Support-Vector Networks , 1995, Machine Learning.

[2]  T.M.A. Maksoud,et al.  Applications of Artificial Intelligence to Grinding Operations via Neural Networks , 2003 .

[3]  Francis Eng Hock Tay,et al.  ε-Descending Support Vector Machines for Financial Time Series Forecasting , 2002, Neural Processing Letters.

[4]  H. M. Khalil,et al.  Support Vector Machines Based Adaptive Controller for Piston Hydraulic Motor , 2011 .

[5]  Sami Ekici,et al.  Support vector machines models for surface roughness prediction in CNC turning of AISI 304 austenitic stainless steel , 2012, J. Intell. Manuf..

[6]  Gerhard Hammann.  Modellierung des Abtragsverhaltens elastischer Schleifwerkzeuge , 1998 .

[7]  Jun-ichi Takahashi,et al.  Vector-field-smoothed Bayesian learning for incremental speaker adaptation , 1995, 1995 International Conference on Acoustics, Speech, and Signal Processing.

[8]  Francis Eng Hock Tay,et al.  epsilon-Descending Support Vector Machines for Financial Time Series Forecasting , 2000, IDEAL.

[9]  Xiang Zhang,et al.  A local process model for simulation of robotic belt grinding , 2007 .

[10]  Davood Afshari,et al.  Creep feed grinding optimization by an integrated GA-NN system , 2010, J. Intell. Manuf..

[11]  Hans Kurt Tönshoff,et al.  Application of fast Haar transform and concurrent learning to tool-breakage detection in milling , 2003 .

[12]  Bernd Kuhlenkoetter,et al.  Real-Time Simulation of Robot Controlled Belt Grinding Processes of Sculptured Surfaces , 2004 .

[13]  Yunquan Sun,et al.  Development of a unified flexible grinding process , 2004 .

[14]  Jianjun Wang,et al.  Process Modeling of Flexible Robotic Grinding , 2003 .

[15]  Qingsong Xu,et al.  Rate-Dependent Hysteresis Modeling and Control of a Piezostage Using Online Support Vector Machine and Relevance Vector Machine , 2012, IEEE Transactions on Industrial Electronics.

[16]  Chen Xiao-qi,et al.  Development of Robotic System for 3D Profile Grinding and Polishing , 2002 .

[17]  Bernd Kuhlenkötter,et al.  Real-time simulation and visualization of robotic belt grinding processes , 2008 .

[18]  XiaoQi Chen,et al.  Robotic grinding and polishing for turbine-vane overhaul , 2002 .

[19]  Francis Eng Hock Tay,et al.  Modified support vector machines in financial time series forecasting , 2002, Neurocomputing.