Application of experimental-based modeling to laser cladding

This article addresses experimental-based modeling of the powder injection laser cladding process. An accurate model for the process is essential for controller design and process automation. There are few dynamic and static models for the process, however, due to the limitations and intensive numerical calculations, their application to real-time process control are not practical. Two model structures, the Hammerstein–Wiener model and Elman recurrent neural network are implemented to identify the laser cladding dynamic model. The type of experiments and collected data are explained, and model predictions for different and unseen data are compared with experimental results. It is shown that the Hammerstein–Wiener approach can more accurately describe the transient response of the process. Results are promising and show that experimental-based models can effectively be used in laser cladding processes.

[1]  Lennart Ljung,et al.  Nonlinear black-box modeling in system identification: a unified overview , 1995, Autom..

[2]  S. Billings,et al.  Identification of nonlinear systems using the Wiener model , 1977 .

[3]  Marco Bianco,et al.  On-line surface treatment feasibility of industrial components by means of CO2 power laser , 1994, Other Conferences.

[4]  G. Bruck High-Power Laser Beam Cladding , 1987 .

[5]  W. Kaplan,et al.  Calculation of process parameters for laser alloying and cladding , 1998 .

[6]  Er-Wei Bai An optimal two-stage identification algorithm for Hammerstein-Wiener nonlinear systems , 1998, Autom..

[7]  Pierre Roussel-Ragot,et al.  Training recurrent neural networks: why and how? An illustration in dynamical process modeling , 1994, IEEE Trans. Neural Networks.

[8]  J. Jouvard,et al.  Continuous wave Nd:YAG laser cladding modeling: A physical study of track creation during low power processing , 1997 .

[9]  J. Mazumder,et al.  One‐dimensional diffusion model for extended solid solution in laser cladding , 1987 .

[10]  Amir F. Atiya,et al.  Multi-step-ahead prediction using dynamic recurrent neural networks , 2000, Neural Networks.

[11]  W. R. Cluett,et al.  Identification of Wiener-type nonlinear systems in a noisy environment , 1997 .

[12]  D. Kechemair,et al.  A Systematic Method for the Design of a Multivariable Controller Actuating Power and Speed during a CO2 Laser Surface Treatment , 1992 .

[13]  Bernardus Engelina Römer Gerardus Richardus,et al.  Process control of laser surface alloying , 1998 .

[14]  Soumitra Das,et al.  Noisy recurrent neural networks: the continuous-time case , 1998, IEEE Trans. Neural Networks.

[15]  Jehnming Lin,et al.  Concentration mode of the powder stream in coaxial laser cladding , 1999 .

[16]  M. Rappaz,et al.  A simple but realistic model for laser cladding , 1994 .