Welcome to the new IOA website! Please reset your password to access your account.

Volume : 44

Part : 2

Proceedings of the Institute of Acoustics

 

 

Subjective evaluation of the acoustic annoyance in a large passenger aircraft cabin

 

Bingcong Lv, Shanghai Jiao Tong University, Shanghai, China
Yu Huang1, Shanghai Jiao Tong University, Shanghai, China
Weikang Jiang, Shanghai Jiao Tong University, Shanghai, China

 

ABSTRACT

 

People demand better acoustic comfort for higher travelling quality and security in aircraft. It is necessary to evaluate and predict the subjective annoyance caused by the noise in aircraft cabins. This study investigates the noise-induced annoyance in a large passenger aircraft. We recorded the noise at 21 positions in the aircraft cabin without passengers and selected 21 stimuli during the cruising. Each stimulus has a duration of 5 seconds and a sound pressure level in the range of 72–81dB(A). The psychoacoustic parameters such as loudness, sharpness, roughness, and fluctuation strength were also calculated. Twenty-four subjects evaluated the subjective annoyance of the stimuli by the absolute magnitude estimation method. Results showed the noise annoyance at the middle section of the cabin is significantly higher than that in the front or rear section. The principal component analysis and correlation analysis found that annoyance is mainly affected by loudness, sharpness, and roughness and is dominated by loudness. We proposed a multiple linear regression model between the subjective magnitudes of annoyance and the psychoacoustic parameters to evaluate and predict the noise annoyance in the aircraft cabin well.

 

Keywords: Noise annoyance, Aircraft cabin noise, Psychoacoustic parameters, Sound quality, Noise evaluate

 

1. INTRODUCTION

 

Acoustic comfort, as the perceived state of well-being and satisfaction with the acoustic conditions in an environment, has attracted more and more attention in the aerospace industry [1-3]. Noise significantly impacts subjective annoyance and increases the awareness of swollen feet, headache, and tiredness, whereas good acoustic comfort diminishes fatigue of the journey, and improves the safety and performance of all occupants [4]. It is essential to investigate, estimate, and design acoustic comfort in large passenger aircraft cabins.

 

There are comprehensive studies on acoustic comfort, including rating noise-induced annoyance with a strong positive correlation with discomfort, unpleasantness, disgust, and disturbance [5, 6]. The sound quality metrics (SQMs), e.g. loudness, sharpness, roughness, and fluctuation strength can accurately represent subjective responses to different sound features [7] and work as psychoacoustic indices for establishing annoyance models, e.g. [7-13]. A few studies have been concerned about

comfort related to noise in the aircraft cabin. Quehl has found increasing loudness, roughness, fluctuation strength, and vibration magnitude would reduce comfort in the aircraft cabin [5]. Both noise level and exposure duration would greatly influence the annoyance [4][14]. Pennig et al. have found the comfort sensation was dominated by sound pressure level (SPL) and significantly affected by the frequency characteristics determined by seat positions in the aircraft cabin (front, middle, rear) [15]. The noise annoyance and its influential factors are complex. Studying the noise annoyance and SQMs may provide a better understanding of subjective responses to noise, and better means to establish quantitative comfort models based on SQMs in the aircraft.

 

This article intends to investigate annoyance caused by noise in a large passenger aircraft cabin during its cruise through a subjective evaluation experiment. The primary purpose is to quantify the subjective annoyance at different positions of the aircraft cabin by the SQMs. Two main hypotheses are proposed: (1) the annoyance differs significantly among different positions of the cabin; (2) the annoyance significantly increased with increasing loudness.

 

2. METHODS

 

2.1. Apparatus

 

The experiment was conducted in the semi-anechoic chamber. The stimuli were played via headphones (Sennheiser HD600, Germany) driven by a digital to analogue converter (RME ADI-2 DAC, Germany) connected to a laptop (Xiaomi Notebook Pro 15, China). The MATLAB (version R2020a) performed as the playing software. The subjects sat on a chair with the headphones in the chamber. The connection diagram of the experiment set-up and the scene during the experiment are shown in Figure 1.

 

 

Figure 1: The scene of the set-up when calibrating the stimuli (left), and the connection diagram of the experiment set-up and the scene during the experiment (right).

 

2.2. Stimuli

 

We recorded noise during a flight of a narrow-body large passenger aeroplane with two jet engines. It was 37.6m long, 11.76m high, 35.8m wingspan, 3.70m cabin width and with a passenger capacity of 150 to 186. The positions of all the measuring points are shown in Figure 1. There were 21 measuring points in the front, middle, and rear sections of the aircraft cabin, as indexed as No 2 ~22 in Figure 2. The microphones of No.2 ~ No.12 were located at the seat headrest with a height of about 1.2m; No.13 ~No.18 at the aisle, with a height of about 1.7m from the floor; No.21 ~ No.24 at about 10cm far from the Air conditioning outlet. All stimuli of this experiment were recorded during the cruising flight. During the recording, the seats back were upright, and the pressurization and air conditioning systems were in regular operation.

We selected 5-s segments from each of 21 noise samples as test stimuli. The stimuli had no voice (e.g. speaking, broadcasting) or sudden abnormal sound (e.g. door opening sound). We calibrated all stimuli via headphones (HD600, Sennheiser, Germany) on a dummy head (HMS IV, Head Acoustics, Germany) to ensure they had the same A-weighted SPLs as the recording samples.

 

 

Figure 2: The measuring points at the front, middle and rear sections of the aircraft cabin.

 

2.3. Subjects

 

Twenty-four healthy subjects, 12 males and 12 females, aged 20 to 30 (median age: 23) yrs, at Shanghai Jiao Tong University participated in the experiment. All subjects had a normal auditory system, no symptoms of ear disease, no blockage of the ear canal, no history of excessive noise exposure, no ototoxic drugs or family hearing loss, and no symptoms of cold or fever during the test.

All subjects signed informed consent before the experiment.

 

2.4. Procedures

 

Twenty-four subjects evaluated the 21 stimuli according to a random play sequence. Each subject experienced stimuli and evaluated the annoyance using the absolute magnitude estimation (AME) method, i.e. using any positive numerical value to rate the annoyance [16]. The value “0” meant ‘no annoyance’, and there was no upper limit for the annoyance value. Before the experiment started, subjects were provided with instructions, and the experimenter would inform them of the purpose and nature of the experiment clearly. All stimuli would be presented to the subjects in random order and could be replayed if required by the subjects.

 

2.5. Data processing


Numerical annoyance values of each subject should be “standardized” or “equalized” because all the subjective magnitudes had to be compared in the same coordinate system. The annoyance magnitudes were divided by the median values of one’s evaluations over all stimuli and then multiplied by 100 [17].

 

Not all of the data sets conform to the normal distribution, so the nonparametric statistical tests were employed. The Friedman test was used to test multiple related data, the Kruskal-Wallis test for testing multiple independent data, and Willcoxon matched signed ranks test for the post hoc test of two groups of related data [18]. Analysis of variance (ANOVA) was used to test the linear regression model.

 

The A-weighted sound pressure level and SQMs, e.g. loudness, sharpness, roughness, and fluctuation strength of all stimuli, were calculated by the Artemis (version 12.1, Head Acoustics, Germany). The multiple linear regression method was used to investigate the relation between the subjective annoyance values and the SQMs. The principal component analysis method and Pearson correlation analysis were used to screen and eliminate redundant independent variables. All the statistical analyses were completed using the IBM SPSS Statistics (version 23, USA).

 

3. RESULTS

 

3.1. Annoyance values and differences in annoyance at different positions

 

All annoyance values quantified by the subjective evaluation experiment are shown in the Box-and Whisker plot (Figure 3). Fourteen abnormal values out of the upper and lower quartile were eliminated.

 

 

Figure 3: The Box-and-Whisker plot of subjective annoyance of aircraft cabin noise. sitting position near the window, standing position in the aisle,  air conditioning outlet.

 

Figure 4 shows the Box-and-Whisker plot of the standardized subjective annoyance values of all subjects. The noise annoyance differed significantly among different positions (p < 0.0001, Friedman). Figure 5 shows the annoyance values of stimuli at three groups of measuring points (i.e., the sitting position near the window, standing position in the aisle, and the air conditioning outlet), respectively. There were significant differences in the noise annoyance of the same group among the front, middle and rear sections of the aircraft cabin (p < 0.0001, Friedman).

 

 

Figure 4: The Box-and-Whisker plot of the subjective annoyance values at different positions of the aircraft cabin. sitting position near the window, standing position in the aisle,  air conditioning outlet. **** p < 0.0001, Friedman test.

 

From Figure 5, there were significantly difference in noise annoyance of the same measuring group among the front, middle and rear sections (p < 0.01, Friedman). The annoyance in the front section was smallest and that in the middle was the greatest for the sitting position near the window group and standing position in the aisle group, whereas the annoyance in the front section was greatest and that in the middle was the smallest for the air conditioning outlet group.

 

 

Figure 5: The Box-and-Whisker plot of the subjective annoyance values of three measuring position groups in different sections of the aircraft cabin. front section, middle section, rear section. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, Friedman test and Wilcoxon test.

 

3.2. The objective acoustic parameters

 

Table 1 lists the A-weighted SPL, loudness, sharpness, roughness, fluctuation strength, and median annoyance values of all 21 stimuli. The loudness calculation was based on the ISO532 method [19], the sharpness was based on the Bismarck sharpness model, the roughness and fluctuation strength were both calculated based on Zwicker models, respectively [7].

 

Table 1: The values of the subjective annoyance, A-weighted SPL, loudness, sharpness, roughness, and fluctuation strength

 

 

3.3. Principal component analysis and Pearson correlation analysis

 

Table 2 shows the component matrix and cumulative contribution rate of the first two principal components from the principal component analysis (PCA). It seems the SPL, loudness, sharpness and roughness have a major contribution to the annoyance.

 

The annoyance, A-weighted SPL, loudness, sharpness, roughness, and fluctuation strength were analyzed by Pearson correlation analysis. The correlation coefficient tables between any two of the five variables of the aircraft are shown in Table 3. There was a strong correlation between annoyance and A-weighted SPL, loudness, and roughness.

 

3.4. The aircraft cabin noise annoyance evaluation model

 

Three SQMs with a high contribution in PCA and high correlation with annoyance in Pearson correlation analysis, i.e. loudness, sharpness, and roughness, were selected as independent variables with the dependent variable, annoyance, for the linear regression analysis. The multiple linear

 

Table 2: The component matrix and cumulative contribution rate of the first two principal components from the PCA.

 

 

Table 3: Pearson correlation coefficient between any two of the five variables of the aeroplane

 

 

regression equation was obtained with high significance (p < 0.001, ANOVA):

 

Annoyance = 36.972 + 7.360Loudness + 54.352Sharpness-103.176Roughness + ε, (1) where the goodness of fit, r 2, was 0.878. The Pearson correlation coefficient between measured annoyance and predicted annoyance reached 0.937, as shown in Figure 6.

 

 

Figure 6: The comparation between the measured annoyance and predicted annoyance


 

4. DISCUSSION

 

In this study, the noise annoyance in the middle section is significantly greater than in the front or rear section for the sitting position near the window (Figure 5). However, Pennig et al. found that noise at the rear section made subjects feel more pleasant and acceptable than in the front or middle section by subjective estimation of short-haul flights noise [15]. It seems the noise characteristics vary in different aircraft types, leading to the different annoyance at different positions of the aircraft cabins. It merits doing more specific investigation on different aircraft types in further study.

 

The negative correlation between annoyance and roughness in the linear regression model, i.e. Eq. (1), merits careful consideration. Similar phenomenons were observed in the previous studies, e.g. on the noise pleasant of vehicle HAVC system [20] and the interior noise discomfort of the micro commercial vehicles [13]. The possible reason is the specific noise characteristics or the multicollinearity in variables of the multiple linear regression. Even so, the proposed model could accurately predict the noise annoyance in the large passenger aircraft cabin.

 

We adopted the simple linear regression between annoyance and each SQM, and the results are shown in Figure 7. The relatively high value of the goodness of fit between annoyance and loudness indicates the dominance of loudness. It is consistent with the previous finding that loudness is the dominant sensation governing annoyance [21]. Although loudness is the dominant parameter affecting subjective annoyance, the changing trend of annoyance does not entirely follow the loudness. The effect of other parameters might be “masked” by the loudness. We need to hold the loudness at the same level to study the influence of other psychoacoustic parameters on annoyance in future study.

 

 

Figure 7: The results of simple linear regression analyses

 

5. CONCLUSION

 

We investigated the annoyance and SQMs of noise in a large passenger aircraft during the cruising flight. The noise annoyance was dominated by loudness, and differed at different positions (i.e., front, middle, and rear) of the cabin. A multiple linear model was proposed to evaluate the noise annoyance by independent variables of loudness, sharpness, and roughness.

 

6. ACKNOWLEDGEMENTS

 

We gratefully acknowledge National Natural Science Foundation of China: 52072242 and Natural Science Foundation of Shanghai, China: 19ZR1424500.

 

7. REFERENCES

 

  1. Airbus, Passengers willing to invest time and money in search of comfort [N/OL], https://www.airbus.com/newsroom/press-releases/en/2013/11/passengers-willingto-invest-time and-money-in-search-of-comfort.html.
  2. Bouwens, J. Ranking of human senses in relation to different in-flight activities contributing to the comfort experience of airplane passengers. International journal of aviation, Aeronautics, and Aerospace, 5(2), https://doi.org/10.15394/ijaaa.2018.1228 (2018).
  3. Vink, P., Bazley, C. & Kamp, I., & Blok, M. Possibilities to improve the aircraft interior comfort experience. Applied Ergonomics, 43(2), 354-359, (2012).
  4. Mellert V., Baumann I. & Freese N., et al. Impact of sound and vibration on health, travel comfort and performance of flight attendants and pilots. Aerospace Science and Technology, 12(1), 18-25 (2008).
  5. Quehl J. Comfort studies on aircraft interior sound and vibration. Universität Oldenburg, 2001.
  6. Jeon J. Y., Hong J. Y. & Jang H. S., et al. Speech privacy and annoyance considerations in the acoustic environment of passenger cars of high-speed trains. The Journal of the Acoustical Society of America, 138(6), 3976-84 (2015).
  7. Zwicker E. & Fastl, H. Psychoacoustics: Facts and Models, 3rd Edition, Springer, 2006.
  8. Torija A. J., Li Z. & Chaitanya P. Psychoacoustic modelling of rotor noise. The Journal of the Acoustical Society of America, 151, 1804 (2022); doi: 10.1121/10.0009801
  9. Brizon C. J. da S. & Bauzer M. E. Combining subjective and objective assessments to improve acoustic comfort evaluation of motor cars. Applied Acoustics, 73(9), 913-20 (2012).
  10. Duvigneau F., Liefold S. & Höchstetter M., et al. Analysis of simulated engine sounds using a psychoacoustic model. Journal of Sound and Vibration, 366, 544-55 (2016).
  11. Zhang E., Zhang Q. & Xiao J., et al. Acoustic comfort evaluation modeling and improvement test of a forklift based on rank score comparison and multiple linear regression . Applied Acoustics, 135, 29-36 (2018).
  12. Li H., Chen K. & Wang X., et al. A perceptual dissimilarities based nonlinear sound quality model for range hood noise. The Journal of the Acoustical Society of America, 144(4), 2300-11 (2018).
  13. Li D., Huang Y. The discomfort model of the micro commercial vehicles interior noise based on the sound quality analyses. Applied Acoustics, 132, 223-31 (2018).
  14. Huang Y., Jiang W. The effect of exposure duration on the subjective discomfort of aircraft cabin noise. Ergonomics, 60(1), 18-25 (2017).
  15. Pennig S., Quehl J. & Rolny V. Effects of aircraft cabin noise on passenger comfort [J]. Ergonomics, 55(10), 1252-65 (2012).
  16. Moskowitz H. R. Magnitude estimation: notes on what, how, when, and why to use it [J]. Journal of Food Quality, 1(3), 195-227 (1977).
  17. Huang Y., Griffin M. J. The relative discomfort of noise and vibration: effects of stimulus duration. Ergonomics, 57(8), 1244-55 (2014).
  18. Siegel S. Castellan. Nonparametric statistics for the social sciences. New York: McGraw-Hill, 1988.
  19. ISO 532-1, Acoustics—Methods for Calculating Loudness—Part 1: Zwicker Method, 2017 Edition.
  20. YOON J-H., YANG I-H. & JEONG J-E., et al. Reliability improvement of a sound quality index for a vehicle HVAC system using a regression and neural network model. Applied Acoustics, 73(11), 1099-103 (2012).
  21. Angerer J. R., Mccurdy D. A. & Erickson R. A. Development of an annoyance model based upon elementary auditory sensations for steady-state aircraft interior noise containing tonal components. No. NASA-TM-104147, 1991.

 

1 Corresponding email address: yu_huang@sjtu.edu.cn