Feedforward and Modal Control for a Multi Degree of Freedom High Precision Machine

Abstract

High precision industrial machines suffer the presence of vibrations due to several noise sources: ground vibration, acoustic noise, direct force disturbances. In the last years the need of higher processing quality and throughput result in a continuing demand for higher accuracy. Therefore vibration isolation systems became mandatory to satisfy these requests. In general, machine supports are designed for high stiffness to obtain a robust machine alignment with respect to its surroundings. However, in the presence of significant ground vibration levels the support stiffness is commonly sacrificed to reduce their transmission to the payload stage. Efforts to go towards these issues are recorded in several applications and the solutions are different for any particular situation, depending on the nature of the vibration sources, the amount of the disturbances and the machine environment. This chapter focuses on the evaluation of a vibration isolation device on the working cell of a micro-mechanical laser center, using active electromagnetic actuators. The machine is composed by two main parts: a frame support and a payload stage where the laser cutting is performed. The machine potential in terms of accuracy and precision is reduced by the presence of two main vibration sources: the ground and the stage itself. The active device should meet two main goals: the payload vibrations damping and the reduction of the transmissibility of ground disturbances. In this work the phases followed to design, realize and validate the device are illustrated with a particular attention to the mechatronics aspects of the project and to the control strategies. The chapter starts on the description of the common solutions and of the techniques described in literature. The requirements analysis and a trade-off phase on the available opportunities for vibration isolation are described. An analysis of the plant components is reported in the second section along with an exhaustive explanation of a) actuation subsystem consisting in four voice-coils, two per axis; b) sensing subsystem aimed to measure the absolute velocities of the frame support and of the stage are measured by means of eight geophone sensors. The considerations leading to the choice of this sensing system are reported along with the signal conditioning block. The active control is performed with a digital platform based on DSP and FPGA. The core of the chapter is the description of the modeling approach and of the control strategies design. The bond-graph approach is used to represent the system behavior, in particular the interactions between the mechanical and electrical subsystems are illustrated. The realized model includes the plant, the sensing, the control and the actuation blocks. The plant is considered as a classical two mass-spring-damper system resulting on a multi-input multi-output system (MIMO), considering disturbances from the stage and the ground and the actuators action between the two masses. Time and frequency domain computations are carried out from the model to evaluate vibration levels and displacements and to identify which parameters need to be carefully designed to satisfy the requirements. The control strategy is focused on the attenuation of the effects of microvibrations on the stage caused by different sources. The technique consists in a combination of two actions, the goal being the minimization of the ground vibrations transmission and the payload vibrations damping: • A single-axis decentralized action consisting in a modal controller used to compensate the high-pass band dynamic of the geophone sensors and to control the vibrations. • A feedforward action working on the disturbances coming from the payload and from the ground. This control is not generated in on-line, but computed in advance from the data of machine responses to the direct disturbances coming from the floor and the stage and resulting in vibrations on the payload and on the frame. The first action itself is aimed to perform active isolation and vibration that nevertheless could be not sufficient for severe specifications applications. The feedforward action is hence used to face this shortcoming by suppressing direct disturbance. The controller design phases along with its performance evaluation are described. The chapter concludes on the illustration of the results obtained with the proposed modeling and control strategy