| A primary concern in rotating machinery is bearing failure. Bearing failure or bearing loss results in damage to other important components of a system and machinery downtime which can be very costly. A common reason for bearing failure, besides improper design and fabrication processes, is dynamic overloading caused by vibration. Active control has recently been applied to control vibrations. In the last decade, displacement amplitude control of rotating machinery has been the generally accepted method, since vibration displacements are easily measured and relatively straightforward to use. When displacement control of rotating machinery is attempted, however, the applied control force can increase the reaction forces developed at the bearings. This is counter to the original intentions of the controller, and can cause severe bearing damage.; This dissertation investigates that problem, and makes the assertion that, in order to most effectively decrease bearing wear, dynamic bearing loads should be controlled as opposed to shaft displacements. Also in this dissertation, two optimal control laws will be developed, one which minimizes shaft displacements and another which minimizes bearing loads. This work compares the two techniques in a flexible beam test rig which represents a typical rotordynamic system.; Designing the controller to minimize shaft displacements is a straightforward approach whereby the LQR performance index is formulated so that modal amplitudes are minimized. Modal amplitudes can be inferred from shaft displacement measurements and used in the feedback control. Designing for bearing load minimization is not as straightforward. Bearings are rarely mounted to measure transmitted forces directly. To circumvent this problem, an output minimization scheme is developed in which bearing loads are written in terms of the shaft displacements, control, and disturbance forces. The loads are then minimized by incorporating those output expressions into the LQR performance index. |
