NEX-LMS: یک طرح کنترل تطبیقی جدید برای کنترل کیفیت صدای هارمونی

This paper presents a novel adaptive control scheme, with improved convergence rate, for the equalization of harmonic disturbances such as engine noise. First, modifications for improving convergence speed of the standard filtered-X LMS control are described. Equalization capabilities are then implemented, allowing the independent tuning of harmonics. Eventually, by providing the desired order vs. engine speed profiles, the pursued sound quality attributes can be achieved. The proposed control scheme is first demonstrated with a simple secondary path model and, then, experimentally validated with the aid of a vehicle mockup which is excited with engine noise. The engine excitation is provided by a real-time sound quality equivalent engine simulator. Stationary and transient engine excitations are used to assess the control performance. The results reveal that the proposed controller is capable of large order-level reductions (up to 30 dB) for stationary excitation, which allows a comfortable margin for equalization. The same holds for slow run-ups (View the MathML source>15s) thanks to the improved convergence rate. This margin, however, gets narrower with shorter run-ups (View the MathML source≤10s).

Interior noise in a vehicle is an important element in the customer perception of the overall vehicle's quality [1], [2], [3], [4], [5], [6], [7], [8] and [9]. The interior noise is made up of contributions from many sources: some only contribute to the overall loudness and annoyance (e.g., uncorrelated road or wind noise), while others reveal important information on the operation of the vehicle and can invoke a desired emotional response [10], as it is the case with engine noise, targeted in this paper. The engine-related interior sound quality design is of major importance for vehicle sound branding, as it underlines its image (e.g., sportiveness, refinement, luxury, etc.) [4] and [11]. In this context, sound branding brings an extra motivation for the use of active control, as it would enable easy and inexpensive adaptations to local markets, of products based on global platforms, meeting distinct customer expectations with the same hardware [4]. Moreover, such control systems can be implemented without compromising other engine/vehicle design attributes, e.g., emissions, fuel consumption, exterior noise, etc. [11]. These two aspects, easy local-marked adaptation and decoupled design parameters, would simplify the NVH (noise vibration and harshness) development process and make it a potential field of application for active control. Demonstrations of the viability of active control in cavity noise applications, including automotive interior noise reduction, have been described by several authors in the past few years [12], [13], [14], [15], [16], [17], [18] and [19]. Usually, the objective of these applications is to reduce the noise generated by the primary source(s) as much as possible. The novelty in this framework is to account for the human perception when defining performance criteria, either to evaluate or to drive the design of active solutions [11], [20], [21], [22] and [23]. In this context, sound quality specialists could prescribe the ideal (or the brand signature) engine sound in terms of its order-level vs. RPM (revolutions per minute) profiles, which would be achieved by means of active sound quality control (ASQC). This paper focuses on the use of ASQC to match prescribed engine order-level vs. RPM profiles. The control strategy is based on the feedforward filtered x-LMS algorithm (Fx-LMS) [24], [25], [26] and [27], to which modifications are proposed in order to improve the convergence rate and allow order-level equalization. This control strategy is described in Section 2. The test setup and the real-time implementation of the controller are described in Section 3. Section 4 presents the experimental results. Finally, some general conclusions are addressed in Section 5.

This paper presents a novel adaptive scheme, NEX-LMS, for the equalization of harmonic noise components with improved convergence. It features a fast converging adaptive algorithm, based on the filtered x-LMS algorithm with normalized reference signal and equalization capabilities, elements necessary for adaptive sound quality controllers. The proposed controller is experimentally validated on a vehicle mock-up which is acoustically excited by a sound quality equivalent engine simulator. The use of such scheme allows repeatable measurements with engine-like excitation signals, furnishing results that can be directly correlated to automotive applications. The results are presented for stationary RPMs as well as for run-ups. They indicate that the controller is effective, reaching up to 38 dB reduction. Also, it was possible to independently control an order, which is an interesting feature for order balancing applications. In order to independently tune multiple orders, similar controllers can be connected in parallel since each narrowband action does not interfere with each other. The improved convergence of the NEX-LMS controller allows to cope with varying engine speeds. The controller has been tested against three run-ups with different lengths (32, 16 and 8 s). As expected, the range of equalization is narrowed for faster run-ups, nevertheless, reductions up to 20 dB can still be observed for the faster run-up. Next steps in this study will address the control of more points inside the cavity, e.g., with a multi-channel adaptive scheme. In addition, the use of the proposed NEX-LMS for controlling roughness will be investigated. For that, not only amplitude but also phase has to be taken into account. The desired order-level profiles (amplitude and phase vs. RPM) could, again, be defined with the aid of sound quality-equivalent models and used to define target values for such a controller.