International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056 Volume: 04 Issue: 08 | Aug -2017 www.irjet.net p-ISSN: 2395-0072 © 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 1441 Real time results of a fuzzy neural network active noise controller Tuan Van Huynh Department of Physics and Computer Science, Faculty of Physics and Engineering Physics, University of Science, Vietnam National University Ho Chi Minh City, Vietnam ------------------------------------------------------------------------------***----------------------------------------------------------------------------- Abstract - The basic principle of activee noise control (ANC) is to create a secondary acoustic noise which has the same amplitude but opposite phase compared with the primary noise in order to attenuate noise in the controlled noise region. This paper presents a fuzzy neural network based filtered-X least-mean-square (LMS) algorithm for ANC system. The saturation of the power amplifier in ANC system is considered. A fuzzy neural network ANC system for compensating the saturation is proposed. An on line dynamic learning algorithm based on the error gradient descent method is carried out. The experimental models of ANC in real time were presented. Key Words: active noise control, adaptive system, fuzzy neural network, real time system. 1. INTRODUCTION Acoustic problems in the environment have gained attention due to the tremendous growth of technology that has led to noisy engines, heavy machinery, pumps, high speed wind buffeting and a myriad other noise sources. Exposure to high decibels of sound proves damaging to humans both in physical and psychological aspects. The problem of controlling the noise level in the environment has been the focus of a tremendous amount of research over the years [1, 2, 7, 9]. Several experiments and simulations are used to demonstrate the various approaches in ANC system. The acoustic and electrical control basis of ANC system is introduced in [1, 2, 3]. Noise cancellation in headphones is introduced in [4]. The filtered-x least mean square (FXLMS) algorithm is a popular adaptive filtering algorithm using a finite impulse response (FIR) filters [1, 2, 7, 12], because it is simple and has relatively low computational load. The development of digital signal processing (DSP) hardware allows more sophisticated algorithms to be implemented in real time to improve the system performance [3, 9, 16]. Linear ANC systems have been successfully used to cancel noise in air conditioning duct systems, handsets, and others [1, 2, 3, 8, 9]. However, in a practical ANC system, the secondary path and primary path of the ANC system may exhibit nonlinear behaviors. The ANC system has to be adaptive because of changes in environment, degradation of system components, and alteration of the noise source. The use of adaptive Volterra filter in ANC system has been presented in [5]. The main drawback of this approach is that the size of the filter increases exponentially with the number of inputs and the computation task is extremely heavy. The use of neural networks has been suggested to cope with the case of nonlinear system [5-8]. The major problem with neural network based ANC system is its relatively slow learning process. In references [9-16] fuzzy-neural and recurrent neural networks have also been used in nonlinear ANC system. Since the fuzzy neural network is a local approximate model, the adaptive process can be accelerated. This paper presents theoretical and experimental modeling of an ANC system in free space by using fuzzy neural network structure. An adaptive feedback ANC system using fuzzy neural network with saturation of the power amplifier is proposed, where the model of fuzzy neural network is simplified to meet the characteristic of an ANC system. Real time identification experiments are performed using a TI6713 floating point DSP board. The applications considered in this paper are headsets, hearing protectors and other assistive hearing devices. The remainder of the paper is organized as follows. Section 2 describe the ANC system and its adaptive algorithm. In section 3, the proposed ANC system is presented. Section 4 demonstrate real time results of the proposed ANC system. The conclusions of the work done as well as suggestions for further research are given in section 5. 2. TRADITIONAL ANC SYSTEM The traditional adaptive feedback ANC system is presented in figure. 1. In figure. 1, the primary noise x(k), generated by the noise source, propagates through the primary path P(z). The secondary noise y(k), generated by the ANC system, propagates through the secondary path S() and G(z) where S() stands for the saturation of the ANC system. The primary noise and the secondary noise are combined to produce the residual noise in the region where the noise is to be controlled. A microphone is placed in this region to measure the residual noise e(k). The fuzzy neural network is used to produce the secondary noise y(k). It is trained such that the residual noise e(k) is minimized. The introduction of the secondary-path transfer function in the system using the LMS algorithm may lead to instability [16]. This is because, it is impossible to compensate for the inherent delay due to G(z) if the primary path P(z) does not contain a delay of equal length. Also, a very large FIR filter would be required to effectively model 1/G(z). This can be solved by placing a model ) ( ˆ z G of the secondary path G(z) in the reference signal path to the weight update of the LMS equation.