Int J Adv Manuf Technol (1996) 11:214-220 © 1996 Springer-Verlag London Limited The International Journal of Rdvanced manufacturinu Technologu De-Skilling Industrial Laser Process Development Ming Y. Huang, Hussein A. Abdullah and Chris R. Chatwin Manufacturing Systems and Informatics Research Group, Department of Mechanical Engineering, University of Glasgow, Glasgow, Scotland A knowledge-based adaptive control environment is constructed to enhance the cost performance of a laser manufacturing system, this decreases the dependence on a skilled technician and reduces process development lead time. The control environment consists of two shells: a pre-process knowledge- based management system (KBMS) expert shell and an adaptive process controller. Experimental results are presented which show how effectively the adaptive control system improves process quality. Keywords: Adaptive control; Knowledge based systems; Laser materials processing 1. Introduction The laser cutting process is a highly nonlinear process and consequently very difficult to analyse or predict analytically. Many researchers have investigated laser cutting phenomena in order to establish an appropriate cutting model to describe and control the laser cutting process. Based on an energy balance method, a simple cutting model was proposed by Chryssolouris and Choi [1] and Steen [2]. Using a cylindrical source model, relationships between the incident power density per unit thickness and the laser beam feedrate in terms of the thermal properties of the target material were established by Bunting and Cornfield [3]. Using stability analysis of the melt flow over the transient cut surface, a thin film flow cutting model was presented by Schuocker [4], Vicanek et al. [5] and Tsai and Weng [6]. Through the investigation of the dynamic behaviour of these thin film flow patterns they characterised the thickness of the thin layer and predicted the striation frequency range for stable cutting processes. Many models of laser cutting processes have been developed [7-9]. However, formulae and data generated by one researcher are often not of use to other researchers as these models are frequently system and set-up dependent. Correspondence and offprint requests to: Professor C. R. Chatwin, School of Engineering, ENGG1, University of Sussex, Falmer, Brighton BN1 9QT, UK. A hierarchically-structured control algorithm that integrates a knowledge-based expert shell and an adaptive process controller has been developed to provide a system-independent and a set-up-unrelated environment for controlling the laser cutting process. Knowledge of laser cutting is organised and exploited using a rule-based system for process optimisation. Cutting feedrate and stand-off height are optimised through an adaptive controller by monitoring the magnitude of the irradiance emitted from the cut front. 2. Knowledge-Based Adaptive Control Environment: System Description A knowledge-based adaptive control environment is con- structed to enhance the cost performance of 1.2 kW Ferranti MFKP CO2 laser manufacturing system, 10.6/xm wavelength. This control environment decreases the dependence on a skilled technician and reduces the time taken in finding the optimal laser operating parameters. The control environment consists of two shells: a pre-process KBMS expert shell and an adaptive process controller. The process knowledge and initial optimal operating para- meters are stored in the knowledge base. A rule-based backward chaining inference engine and least-square parameter estimation are used to relate material selection to operating parameter evaluation. When a cutting query for a particular material is made, the initial optimal operating parameters are deduced from the knowledge-based expert shell and fed back by the operator via the graphical user interface (GUI). When the cutting process is initiated, the adaptive controller is activated. Optimum operating parameters, reasoned from the expert shell, are transferred to the control actuator unit. Settings of laser power mode (i.e. pulsed or CW) and power range are then downloaded to the laser pulser unit. Settings of cutting feedrate and stand-off height are also downloaded to the Heidenhain TNC controller unit. The cutting process is then initiated by the TNC controller unit. Irradiance emitted from the cut front during the laser cutting process is sampled, by a photodiode, and fed back to the process adaptive controller which provides optimal control of the cutting process based on this irradiance measurement. The system diagram is illustrated in Fig. 1.