Surface Roughness Control Simulation of Turning Processes

The machining quality on computerized numerically controlled (CNC) machine tools is sensitive to the machining parameters. With modern machine tools, an operator still manually adjusts controlling parameters, such as feed rate and cutting speed. The adjusted values mainly depend on operator experience and knowledge. Standard machining catalogues and commercial cutting condition prediction software [1] generally offer only recommended values, which are far from optimum. Usually, technicians set the highest feed rate possible. The limitations are the available machine power, tool strength and surface roughness tolerance. The increase of the cutting force due to the higher feed rates increase the tool and work piece deflection, machine chatter; consequently, work piece precision is reduced. Therefore, maintaining the cutting force on the tool tip at the appropriate value is a method to ensure dimensional accuracy. To solve this problem, it is necessary to modify the controlling parameters on-line [2]. Adaptive control (AC) accomplishes this. The AC of turning operations is a logical continuance of the evolution of CNC systems. The goal of adaptive control is to control the turning process via the on-line adjusting of controlling parameters [3], which are subject to machining constraints. An AC system is introduced in the cutting process by Stute and Goetz [4]. Such systems to maximize the feed rate for a specified cutting force in turning operations have been proposed in the past [5]. Many AC systems that employed on-line adjustment of the feed rate to ensure a stable and efficient turning have been proposed [6] and [7]. Thus far, no control system has been proposed to adjust cutting condition to maintain surface roughness constant in prescribed tolerances. Model reference adaptive control (MRAC) systems have been also developed for turning [8]. MRAC systems, developed from AC theory, are widely used due to their robustness and disturbance rejection capability. The basic adaptive model reference adaptive controller approach was originally investigated by Landers and Ulsoy [9]. These controllers were simulated and evaluated and physically implemented by [10]. Numerous forms of MRAC system have been developed, especially in milling [11] and [12]. Non-deterministic MRAC controllers are not always suitable for real time control [13] because the controller requires an explicit model and significant computation. The neural network-based MRAC systems [14] and [15] allow much faster development of the models and thus make these controllers practical. The main drawback of integrating MRAC systems in turning is that many algorithms generate significant oscillatory behaviour at low depths of cutting [10]. The model-based control system developed in our research overcomes this Surface Roughness Control Simulation of Turning Processes Čuš, F. – Župerl, U. Franci Čuš – Uroš Župerl* University of Maribor, Faculty of Mechanical Engineering, Slovenia