An Improved System of Active Noise Isolation Using a Self-sensing Actuator and Neural Network

Abstract

In this paper we present an improved active noise isolation method, consisting of a self-sensing actuator, a neural network identifier and an adaptive feedback controller using a finite impulse response (FIR) filter and the Filtered-X LMS algorithm, in which no acoustical sensors were necessary to suppress the noise transmission through a plate structure. The structure is a composite plate with an embedded piezoelectric patch. Based on the self-sensing technique, the same piezoelectric element functions as both a sensor and an actuator. A bridge circuit was used to separate the sensor signal from the actuator signal on the piezoelectric patch and the obtained signal was used in the identification of the sound pressure of a point in the space. A neural network was used instead of the Rayleigh’s integral formula for the identification of the sound pressure as used in the former study. The results show that the proposed control approach using both a self-sensing actuator (SSA) and neural network identifier exhibited better noise control performance than using Rayleigh’s integral formula. It also exhibited similar noise control performance to the traditional control system using a microphone, although the new system used only one piezoelectric patch for both the sensor and actuator.