Modelling, Identification and Control of Thermal Deformation of Machine Tool Structures, Part 5: Experimental Verification

Abstract

Machining accuracy is more often governed by thermal deformation of the machine tool structure than by static stiffness and dynamic rigidity. Since thermally induced errors cannot completely be eliminated at the design stage, the use of control and compensation systems is an inevitable course of action. Existing control systems are based on two different approaches; the use of empirical compensation function, and on-line execution of numerical simulation models. To overcome the limitations of these methods, a new control system has recently been proposed by the authors. This system, which is based on the concept of generalized modelling, incorporates a realtime inverse heat conduction problem IHCP solver to estimate the transient thermal load applied to the structure. With this information, the relative thermal deformation between the tool and the workpiece is estimated and used as a feedback control signal. In previous parts of this series, computer simulation test cases were carried out to examine the dynamic response, accuracy and stability of the system. In the present study, the performance of various components of the control system, specifically, the IHCP solver, the thermal deformation estimator, and the feedback controller are verified experimentally using a three-component structure. The results showed that the derived generalized thermoelastic transfer functions and algorithms are indeed quite accurate in predicting and controlling the transient thermoelastic response behaviour of a predominantly linear structure. The results showed that the a IHCP solver is inherently stable even when the temperature measurements are contaminated with random errors. The excellent computational efficiency of the integrated system is shown to be well suited for real-time control applications involving multi-dimensional structures, achieving a control cycle of less than 0.5 second. The experimental results showed that in real structures higher modes can be present, and therefore, a fourth order deformation model should be used to improve the prediction accuracy. The proposed PID control system, with feedforward branches, was capable of reducing thermal deformations of the order of 200 μm to levels below ±8 μm. These results also demonstrated the effectiveness of artificial heat sources as a control actuation mechanism, in spite of their inherent limitations, namely, thermal inertia, coupledness, and unidirectionality.