Author (s)
Marco Favaretto
Affiliation
Polytechnic of Turin (Master of Science Course in Cinema and Media Engineering)
Publication date
2024
Abstract
In the dynamic world of agricultural machinery, tractors are indispensable tools, working long hours in diverse and challenging environments. The well-being and productivity of the operator is closely linked to the comfort and functionality of the cab. As the demand for ergonomic and technologically advanced tractor cabs continues to grow, there is an urgent need to address the sound quality of these environments. In order to improve overall comfort, tractor cabs typically incorporate measures such as sound insulation to minimise vibration and noise caused by the engine. However, existing techniques primarily address engine-related issues and there is currently a lack of research and established methods to effectively mitigate interior noise generated by HVAC systems in tractor cabs. It should be noted that during the summer months, when temperatures are higher, operators are forced to run the HVAC system at maximum speed, resulting in increased noise levels. The expected increase in HVAC system noise levels on hot days is a critical consideration, given the potential impact on operator experience and comfort. This thesis focuses on a predictive modelling approach to evaluate the sound quality parameters associated with the interior noise of a tractor HVAC system, combining objective measurements with subjective evaluations. The methodology begins with an extensive review of the relevant literature on vehicle interior noise prediction models, focusing on which psychoacoustic parameters are used, what type of subjective evaluation is used, and the best performing prediction models. This foundation serves as a critical framework for the subsequent stages of the study, ensuring a comprehensive understanding of the current state of research in this area. Signal acquisition tests are an essential part of the methodology and involve the use of a 19-channel microphone, an artificial head with two types of binaural microphone headsets and an omnidirectional microphone. The recorded cabin noise, captured at the operator’s ear level, serves as the empirical basis for the analysis of the psychoacoustic parameters. Recognising that the sole calculation of sound pressure level (SPL) may not be optimal for assessing noise discomfort, the study goes beyond SPL by meticulously calculating additional psychoacoustic parameters. Parameters such as loudness, A-weighted sound pressure level (A-SPL), sharpness and roughness are included in the analysis to provide a more comprehensive understanding of the perceptual aspects of tractor HVAC interior noise. The evaluation of sound quality is fundamentally based on the subjective perception of the individual. Through meticulous subjective testing, objective parameters are harmoniously integrated with human perceptual experience. This synthesis of empirical data and subjective ratings forms the basis for the construction of predictive models aimed at understanding and quantifying sound quality. Two subjective rating methods are proposed and compared: the first method uses sound judgement with a 1-10 rating scale measuring annoyance, while the second method, called Semantic Differential, uses 7-point scales with paired bipolar adjectives related to loudness, A-SPL, roughness and sharpness. The subjective tests are carried out using both binaural listening with a computer and spatialised audio listening (3rd order ambisonics) with a virtual reality (VR) headset simulating the tractor cab environment. Furthermore, the research endeavors to develop a sound quality prediction model through the application of multiple linear regression. The predictive capabilities of the model are rigorously assessed, with particular aYention to its accuracy in forecasting psychoacoustic parameters and subjective noise ratings. Preliminary findings suggest promising outcomes, with the 1-10 rating scale demonstrating exceptional efficacy in predicting noise annoyance, achieving an impressive R-squared value of 0.97. Additionally, the Semantic Differential Method (SDM) showcases its utility in predicting psychoacoustic parameters, particularly excelling in roughness (R-squared of 0.88) and loudness (R-squared of 0.97). However, challenges are encountered in reliably predicting sharpness, aYributed to significant errors and discrepancies in subjective ratings. The findings of this study form the basis for the development of advanced predictive models, potentially using neural networks. These models will make a significant contribution to the improvement of tractor cabin, promoting a more comfortable and operator-friendly environment in agricultural machinery.
Full paper
https://webthesis.biblio.polito.it/secure/30900/1/tesi.pdf
Keywords
tractor cabin acoustics, HVAC noise, psychoacoustic parameters, sound quality prediction, subjective evaluation, 3D audio, ambisonics, virtual reality simulation