Controle ativo acústico estrutural: projeto, simulação e análise de qualidade sonora

Abstract

This master thesis deals with the study of active noise and sound quality control systems in automotive applications, and the innovative assessment of their performance based on relevant psychoacoustic metrics in this area such as Loudness, Roughness and Tonality. A methodology for designing and implementing active control systems based on psychoacoustic metrics is established. The main challenges reside in defining the set of time-frequency attributes of the disturbance, responsible for the undesirable auditory stimuli. Computer algorithms are used extensively for simulating and analysing the disturbance, which is a synthesized internal combustion engine noise, integrating both the control system design and psychoacoustics areas. An adaptive active sound quality control algorithm is proposed to resolve the amplitude and/or relative phase interactions among the narrowband components of the multi-harmonic disturbance. Eventually, the numerical results are correlated with real experiments, allowing not only a critical analysis of the psychoacoustic performance of the controllers, but also of the proposed design methodology, which is accomplished with the aid of a vehicle's mock-up, subject to disturbances of relevant psychoacoustic characteristics to the proposed problem.

Sound quality driven active control of periodic disturbance for hybrid vehicles

“…The possibility for tailoring the residual powertrain noise to provide the driver and passengers with relevant auditory information on the operation regime of the vehicle (KUO et al, 1997;DIEGO et al, 2004;GONZÁLEZ et al, 2006;KUO et al, 2007;WANG et al, 2010), or even for enhancing customer perceptions (GONZÁLEZ et al, 2003;REES;ELLIOTT, 2006) of, e.g. vehicle quality, powerfulness, sportiveness, luxury, reliability, quietness, among others (JAY, 2007), also allows the manufacturers to boost their products with unique and distinctive sound marks (FASTL, 2005;KUO et al, 2008;OLIVEIRA et al, 2010;CHEONG, 2010;WANG et al, 2013;MOSQUERA-SÁNCHEZ;OLIVEIRA, 2014b). At this point, SQ techniques come in handy for supporting auditoryoriented procedures, since they bridge the gap between objective vibro-acoustics and the human auditory response, by means of models that objectively quantify auditory perceptions.…”

“…Whereas it is still possible to cope with the unusual amount of harmonics and spectral crossings amongst them in the hybrid electric powertrain noise, by merely following SQ techniques applied for ICE noise, the fact that its spectrum spreads over a larger bandwidth, sometimes in excess of 4 kHz, does demand the inclusion of more auditory dimensions to the problem, namely other than the widely used low-frequency Loudness (GONZÁLEZ et al, 2003;JEON et al, 2007;LEE, 2008;OLIVEIRA et al, 2009;NYKÄNEN;SIRKKA, 2009;SHIN et al, 2009;CHEONG, 2010;CHO et al, 2011;TAN et al, 2012;MOSQUERA-SÁNCHEZ;OLIVEIRA, 2014b) and Roughness (GONZÁLEZ et al, 2003;JANSSENS et al, 2007;JEON et al, 2007;LEE, 2008;OLIVEIRA et al, 2009;LEE et al, 2010;CHEONG, 2010;CHO et al, 2011;WANG et al, 2013;MOSQUERA-SÁNCHEZ;OLIVEIRA, 2014b;DRAGONETTI et al, 2017). In fact, some recent works on NVH aspects of HEVs (GOVINDSWAMY et al, 2009;MOLLET et al, 2015;GUO et al, 2015) put forward the need for developing SQ approaches that face tonal, narrowband and sharp sounds, which are the novel high-frequency occurrences in the noise that come with integration of the EM in the powertrain.…”

“…Since the objective of the control actions is to independently profile a periodic sound at a number of sensor locations, by means of either balancing the amplitudes, or shifting the relative phases of the harmonics, the kind of multichannel algorithm that minimizes independent cost functions becomes of interest. Hence, based on the demonstrated effectiveness of SISO systems in profiling periodic sounds (REES; ELLIOTT, 2006;KUO et al, 2007;KUO et al, 2008;OLIVEIRA et al, 2009;OLIVEIRA et al, 2010;MOSQUERA-SÁNCHEZ;OLIVEIRA, 2014b), the proposed approach to the problem consists on placing a single sensor at each of the M targeted locations, together with as many actuators as sensors. Such an arrangement leads to MˆM secondary paths, from which only M secondary paths should be identified, according to the current decentralization strategy for multichannel control schemes, illustrated in Fig.…”