Psychoakustische Bewertung verbrennungsmotorischer Geräusche

Abstract

Beim Kauf eines Pkw wird der akustische Qualitätseindruck eines Fahrzeugantriebstrangs ste-tig relevanter. Die wahrgenommene Geräuschqualität des Motoraggregates nimmt hierbei eine Schlüsselrolle ein. Aufgrund der Ausprägung einzelner Störgeräusche wird jene Geräusch-qualität negativ beeinflusst. Diese als unangenehm empfundenen Geräuschkomponenten gilt es im Rahmen der Fahrzeugentwicklung weiter zu reduzieren, wobei die Identifikation und Bewertung störender Geräuschanteile am Motorgesamtgeräusch Voraussetzungen für eine effektive Akustikoptimierung sind. Vor allem die impulshaften Motorstörgeräusche „Nageln“ und „Tickern“ werden bei Diesel- bzw. Otto-DI-Motoren als besonders lästig eingestuft, wes-wegen diese Arbeit darauf abzielt, die nagelnden sowie tickernden Geräuschanteile aus dem Gesamtgeräusch zu analysieren und zu bewerten. Hierzu wurden empirische Formeln entwi-ckelt, welche jene Geräuschanteile in ihrer Intensität in Grenzen einstufen können. Diese ist rein empfindungsbasiert und besteht aus der Impulshaftigkeit, der Lautheit, der Rauigkeit so-wie der Schärfe des Motorgesamtgeräusches. Wie auch bei anderen psychoakustischen Be-wertungsskalen wurde die Benotung von 1 (sehr nagelnd/tickernd) bis 10 (nicht nagelnd/ti-ckernd) vorgenommen. Anhand von Hörversuchen mithilfe von Expertenjurys konnten die Die-selnagelgeräusch- und Tickergeräuschbewertungsformeln anschließend verifiziert werden. Hiernach kann präzise ausgesagt werden, in welchen Motorkennfeldbereichen die Nagel- so-wie Tickergeräusche die Angenehmheit des Motorgesamtgeräusches negativ beeinflussen. Ferner wurde die Dieselnote im Rahmen einer körperschallbasierten Dieselmotorenregelung angewendet. Es wurde untersucht, welche Verbesserungen sich durch die Integration geeig-neter akustischer Sensorsignale in das Motormanagement bezüglich des Geräusches erzielen lassen. Hierbei konnten Korrelationen zwischen Körperschall- und Zylinderdrucksignalen mit-hilfe der Kohärenz und der Wigner-Ville-Transformation aufgezeigt werden. Mit der anschlie-ßenden Merkmalsextraktion durch die digitale Verarbeitung der Körperschallsignale war es möglich, Merkmale der Verbrennung und der Einspritzung zu entnehmen, welche im Motor-management eingesetzt werden konnten. Durch Regressionsmodelle konnten somit Verbren-nungsgrößen ermittelt werden, mit denen der indizierte Mitteldruck und die Verbrennungs-schwerpunktlage in einem Motormanagementsystem selektiv regelbar sind. Damit können günstigere und robustere Körperschallsensoren die kostenintensiven Zylinderdrucksensoren ersetzen. Des Weiteren ermöglichte die hohe Korrelation zwischen Luftschall und Körperschall eine Bestimmung der Dieselnote auf Basis der Modulation und der Lautheit, welche aus dem Körperschallsignalen abgeschätzt werden konnten. Die Information über die Verbrennungsge-räusche in Form der Dieselnote konnte dadurch ebenfalls erfolgreich in das Motormanagement integriert werden. Zur Optimierung der Verbrennungsgeräusche wurde der Einfluss der Voreinspritzung vornehmlich untersucht. Es konnte durch einen heuristischen Optimierungsalgo-rithmus eine arbeitsspielsynchrone Adaption der Voreinspritzparameter erzielt werden, um die Lästigkeit der Verbrennungsgeräusche auch im transienten Betrieb gezielt zu mindern.

The customer's choice when buying a car is more and more driven by acoustic quality impres-sion. The perceived sound quality of the engine unit plays a key role here. Due to the nature of individual background noises, the sound quality is negatively influenced. These noise com-ponents, which are perceived as unpleasant, need to be further reduced in the course of vehi-cle development, whereby the identification and evaluation of disruptive noise components in the overall engine noise are prerequisites for effective acoustics optimization. Especially, the impulsive engine noises "knocking" and "ticking" are classified as particularly annoying in Die-sel and direct-injection Otto engines. So, the main focus of this thesis aims to analyze and evaluate the knocking and ticking noise components from the overall engine noise. For this purpose, empirical formulas were developed which are able to classify those noise compo-nents in their intensity within chosen limits. On this behalf, a preselection of perception-based variables was made to determine, which would be suitable for the respective formula approach. The choice fell on the impulsiveness, the loudness and the roughness for the Diesel grade as well as the impulsiveness, the loudness and the sharpness for the ticker grade. For the ticker noise analysis, the Injection Analyzer (injection test bench) was used, by which the individual noise components in the injector could be determined both in time and in fre-quency domain. The Smoothed-Pseudo-Wigner-Ville distribution proved to be an effective tool for analyzing the separate noise phenomena during the injection and combustion processes. This made it possible to detect the ticker noise as precisely as possible in airborne and struc-ture-borne noise. Based on this, the psychoacoustic variables were eventually combined in the context of empirical formulas, given experimentally determined weighting coefficients and iter-atively adjusted further on. After the completion of the formulas, they were validated with the help of auditory experiments, whereby a sufficiently good correlation resulted between the em-pirical formulas and the subjective perception of the test subjects. As with other psychoacoustic evaluation scales, the rating was made from 1 (very knocking/ticking) to 10 (not knocking/tick-ing). According to these results, it can be precisely stated in which engine map areas the knock and ticker noise undermines the pleasantness of the overall engine noise. Another main achievement of the work is the application of the Diesel grade within the frame-work of a cylinder-selective, structure-borne noise-based Diesel engine control. It was investi-gated which improvements can be achieved by integrating suitable acoustic sensor signals in the engine management with regard to the overall engine noise. The aim hereby was initially to determine the combustion characteristics from acceleration sensor signals and control unit variables in order to implement an engine management based on cylinder pressure. In addi-tion, the engine noises were modeled using the structure-borne noise signals to determine the annoyance of the combustion noises and to integrate them into the engine management sys-tem. Thanks to the virtual pressure and virtual noise sensors developed, it was possible to improve the combustion process as well as the noise emissions. These virtual sensors result from the estimation of the respective parameters by means of the structure-borne noise. Hereby, the first step was to record sufficient information about the combustion with structure-borne noise sensors. By means of detailed investigations in the time-frequency domain, it was possible to find dependencies between the cylinder pressure and the structure-borne noise signal, with frequency ranges being selected from which a feature of the combustion angle position in the structure-borne noise signal could be extracted. A correlation analysis between these extracted features from the structure-borne noise signals and combustion angle posi-tions showed good results. For the implementation of the cylinder pressure-based engine con-trol the adjusted variables must be estimated, since the combustion variables cannot be de-rived directly. With the aid of control unit data and the structure-borne noise characteristics, regression models could be created for the cylinder-selective estimation of the center of com-bustion, the maximum cylinder pressure and the indicated mean effective pressure. The Diesel grade, which could also be determined by modulation and loudness of airborne sound signals, was used to evaluate the combustion noise. To estimate the Diesel grade on the basis of the structure-borne noise, the idea was pursued to determine the modulation and loudness directly from this data, whereby a high correlation could be demonstrated. Further-more, the estimation of the airborne sound quantities was implemented using a regression model, in which additional control unit data was used. This was followed by the sequential creation of the engine management, whereby a cylinder-selective control of the indicated mean effective pressure was initially developed, which uses the estimated, structure-borne noise-based model variables as a control variable. Subsequently, a regulation of the center of com-bustion was implemented, which uses the estimated model variables. Individual jump meas-urements were carried out to verify that regulation. The measurements confirmed the success-ful implementation of the control concept based on structure-borne noise-based combustion model variables. The pre-injection parameters were also used to minimize the annoyance of the combustion noise. On the basis of preliminary investigations at two operating points, it was found that the Diesel grade changes significantly in a limited angular position range, especially compared to the pre-injection quantity. To minimize the Diesel knocking, a heuristic optimiza-tion algorithm was able to achieve an iterative adjustment of the pre-injection parameters syn-chronized with the work cycle in order to reduce the annoyance of the combustion noises, even in transient operation.