توسعه الگوریتم تطبیقی ​​برای کنترل فعال کیفیت صدا

This paper develops an active sound-quality control (ASQC) system based on the active noise equalization (ANE) technique, and optimizes it with the filtered-error least mean square (FELMS) algorithm and normalized reference signal generator. The ASQC system controls the sound quality of products such as engines by changing the amplitudes of harmonics. This optimized system uses the FELMS algorithm to limit disturbances in the passband caused by uncorrelated interferences with high gains in the secondary path, thereby increasing the system stability. It achieves fast convergence by normalizing the amplitudes of internally generated sinusoids in the reference signal according to the magnitude response of the secondary path at the corresponding frequencies. Computer simulations demonstrate the desired spectral shaping capability with faster convergence and reduced passband disturbance.

Active noise control (ANC) [1] and [2] uses the principle of superposition, whereby a secondary noise of equal amplitude and opposite phase cancels an unwanted primary noise. In many practical ANC applications, the primary noise produced by rotating machines (such as engines) is periodic and contains multiple harmonic-related narrowband components. In the periodic ANC system, a nonacoustic sensor such as a tachometer or an accelerometer [3] can replace the reference microphone. The sensor output synchronizes an internally generated reference signal, thus preventing feedback from the secondary source to the reference sensor. This periodic ANC system was analyzed in Ref. [4] using the filtered-X least mean square (FXLMS) algorithm [5], where the reference signal is filtered by an estimate of the secondary path for updating the adaptive filter's coefficients. The design of an ANC system usually pursues maximal attenuation of the primary noise. However, in some applications, it is desirable to retain a small residual noise with a specified spectral shape. For example, in an automobile, or earth-moving machine, the driver needs some audible information about engine speed to be able to control the vehicle safely. Active sound-quality control is another potential use of ANC, which changes (amplifies or attenuates with predetermined values) the amplitudes of noise components to improve sound quality or change noise signature. These demands lead to an extension of the ANC concept to include ASQC. The equalization system for periodic noise with multiple harmonics is called a narrowband ANE, which has been developed in Refs. [6] and [7]. This algorithm also has been implemented and analyzed in the frequency domain [8], and extended to equalize broadband noise [9]. The fundamental difference between the narrowband ANC and ANE is the system transfer functions. The magnitude response of the narrowband ANC system presents multiple notches with infinite nulls at controlled frequencies, so that the system attenuates those noise components completely. However, the depths of individual notches of the narrowband ANE system are independently adjustable and can even be changed to peaks with predetermined amplitudes without affecting the characteristics of other notches. Active sound-quality control using the narrowband ANE algorithm inherits the problems of ANC systems with the FXLMS algorithm. These include passband disturbances due to uncorrelated interference at frequencies where the magnitude response of the secondary path has high gain [10], and slow convergence due to the eigenvalue spread of the input autocorrelation matrix determined by the magnitude response of the secondary path [11]. This paper modifies the solutions [10] and [11] for solving these two critical problems in ANC systems, integrates them with the narrowband ANE [6] and [7], and presents the optimized ASQC algorithm [12].

The ASQC system based on the ANE system with the optimized FELMS algorithm has been developed, tested, and analyzed. Simulations show that this new system can shape the residual noise spectrum, thus effectively controlling the sound quality or signature. This new system also reduces passband disturbance, which is caused by uncorrelated noise components in the primary noise that are amplified by the secondary path. Faster convergence can be achieved by using reference sinusoids with normalized amplitudes instead of unity amplitude. Therefore, the ASQC system developed in this paper provides faster convergence with reduced passband disturbance while controlling the sound quality.