Abstract
This paper describes a new control algorithm for reducing sound pressure level of a fan noise. As the algorithm basically comes from the learning control that is invented for controlling the repetitive motion of robot, it is effective for reducing the periodic component of fan noise. The control system utilizing the algorithm includes only error microphones without reference signal sensor, which means the computer program realizing the algorithm is far shorter than that for conventional control algorithm, filtered-X or fitered-ε, for example. In this paper the efficiency of the control matrix, which works as the core of the control and is conditional for positive definiteness, is investigated. The control matrix is induced from a impulse response of the stable transer function. Efficiencies of the matrices are compared by using control simulation for a periodic fan noise.
