Noise Reduction for Structures by Using Magnetic Vibration Absorbers : Application of Neural Networks for Obtaining Optimal Parameters of the Noise Reduction Absorbers

We present a system for reducing noise from structures using electromagnetic vibration absorbers. Conventional design techniques for vibration absorbers are not appropriate for noise reduction absorbers, because higher-mode vibrations exert strong effects on noise. Hence, we first derive an analytical expression for noise power including higher modes, then present a method for obtaining optimal parameters for noise reduction absorbers. In this method, noise power is taken as a function of cost, and the parameters are decided by using a neural network procedure. An improved algorithm of the neural network for obtaining the parameters is given, experimental tests are performed and the theoretical values are compared with the experimental data.