%0 Journal Article
%T Self-Identification Algorithm for the Autonomous Control of Lateral Vibration in Flexible Rotors
%A Thiago Malta Buttini
%A Rodrigo Nicoletti
%J International Journal of Rotating Machinery
%D 2012
%I Hindawi Publishing Corporation
%R 10.1155/2012/873645
%X Intelligent machines are capable of recognizing their operational condition and take actions towards optimality through an autonomous processing of information. Considering the importance of rotating machines in modern industry, this concept of intelligent machines can be applied to achieve high availability, thus avoiding interruptions in the production flow. In this work, a self-identification algorithm is proposed for the autonomous decision and control of a flexible shaft rotating system with electromagnetic actuators. Based on the D-decomposition technique, the algorithm searches in the domain of controller gains the best ones for P and PD controllers to reduce maximum peak response of the shaft. For that, frequency response functions of the system are automatically identified experimentally by the algorithm. It is demonstrated that regions of stable gains can be easily plotted, and the most suitable gains can be found to minimize the resonant peak of the system in an autonomous way, without human intervention. 1. Introduction According to [1], an intelligent machine is capable of recognizing its operational condition and takes actions towards optimality through an internal and autonomous processing of information. This can be understood as a system able to adapt and learn from plant data and assure the desired dynamic behavior of the plant under any disturbance and/or change in its inner characteristics [2]. In the last decades, actuators and sensors have been incorporated into rotating machines aiming at attenuating and controlling the vibration levels, especially those presented by the shaft [3¨C6]. In such cases, the actuators can also be used as exciters, and open-loop frequency response of the system can be identified. Such frequency response would contain information from the actuator system + rotating system/bearings + sensor system that can be used to find the best gains for the controller. Therefore, if all this process (identification of plant FRFs and determination of optimum gains) is automatically managed by an algorithm, as proposed in the present work, one would achieve the definition of intelligent machines stated by [1]: a machine that can identify its characteristics (self-identification) and take actions towards optimality (search for optimum gains). In the area of active magnetic bearings (AMBs), much has been done in this direction by the development of adaptive control systems. In most cases, the conventional control system, responsible for rotor levitation, is complemented by an additional adaptive algorithm, responsible for
%U http://www.hindawi.com/journals/ijrm/2012/873645/