Author (s)
Elena Forneris
Affiliation
Polytechnic of Turin (Master of Science Course in Cinema and Media Engineering)
Publication date
2024
Abstract
In the agricultural world, tractors are an indispensable tool for farmers’ work. The characteristics and
requirements of these machines are increasingly stringent from a technological and performance point of
view, with means capable of facing the most difficult territories and for very long working hours.
From a technological point of view, the tractors have made significant progress. The cabin of agricultural
vehicles has also undergone many technological innovations, both regarding instrumentation and in terms of
comfort and well-being for the user. These improvements, however, are still at a rather late stage as regards
the acoustic comfort inside the tractor cab. From the acoustic point of view, investigations and studies have
been conducted mainly on external noise, the engine and the vibrations, squeaks and internal noises it causes.
The cab is therefore being insulated and there is already an interest in the acoustic comfort of tractor users.
What is not considered, however, is the fact that farmers work in very hot conditions and for extended hours,
which forces them to seek relief by using HVAC, commonly called air conditioning. HVAC is a fundamental
component of every vehicle, as it allows to have a high comfort from the point of view of the temperature. This
is particularly true for agricultural vehicles such as tractors, since the working conditions to which users are
subjected are very heavy from a temperature point of view.
The HVAC block creates a noise level inside the cab, however, which is harmful to farmers who spend hours
on the trucks, as exposure to too high levels of noise for too long are a risk to the physical and psychological
health of users, with short- and long-term damage not to be underestimated.
This thesis, therefore, investigates the acoustic quality inside a tractor, with objective measurements and tests
subjective, the results of which have been combined and used to calculate a linear predictive noise model
same. The research is the continuation of the thesis work carried out by my colleague Marco Favaretto [39].
Many articles and methodologies are the same as those used by him. On the other hand, they have been
differences in methodology and insights at various stages of work. Favaretto’s work will often be cited, both
as a source and for comparison with acquired data, in order to assess whether the differences introduced from
a methodological point of view have been effective. The methodology consists of several parts. The
methodology is multi-part.
Initially, a literature review was conducted on the topic to have a verified methodology, to implement a choice
of sound and to choose an effective prediction model and a correct strategy for conducting subjective tests.
This first part allowed to have a clear picture of the state of art and research in the field.
The second phase consists of the acquisition of signals in the tractor cab, with the various settings studied by
the HVAC block. Three different types of microphones were used for this part, such as a 19-channel
microphone, an omnidirectional sound meter and a binaural microphone consisting of an artificial head. The
measurements were carried out with the engine off, with the tractor located outdoors in a quiet area, with the
cab doors closed, to simulate the actual use and the situation of use. The microphones were therefore placed
at the driver’s seat about the height where the driver’s ears would be. The recordings were essential to extract
the parameters and objective values of the noise produced by the HVAC block. The psycho-acoustic
parameters analyzed and extracted were varied, all supported by literature, such as loudness, sharpness,
roughness, tonality, fluctuation strength, A-weighted sound pressure level and non-weighted sound pressure
level. This allowed a very clear and wide picture of the behavior and type of noise in the cab.
The third stage consists of subjective tests, which are essential to have real human perception of noise. The
tests were conducted with three different methodologies: a first test was carried out on the day of
measurements, with the subjects tested in the tractor cab listening to the noise produced by the HVAC in
person. In a separate location, three replaying tests were carried out, one binaural in headphones, the second
one in the Audio Space Lab in Politecnico of Turin, with a third order ambisonics system of audio reproduction
ad a VR headset, and the other with the aid of a VR headset and a set of headphones reproducing spatialized
audio, where the recordings detected on the ground were used as tracks. In order to limit the statistic variance
of the results, the laboratory tests were repeated by 5 subjects. The tests asked subjects two types of questions:
a scale from 1 to 10 which assessed the annoyance of sound, the other following the method of semantic
differential methods, studied in literature, with semantic scales relating to all the objective psycho-acoustic
parameters studied.
The predictive capabilities of the model are rigorously assessed, with particular attention to its accuracy in
forecasting psychoacoustic parameters and subjective noise ratings. Preliminary findings suggest promising
5 outcomes, with the 1-10 rating scale demonstrating exceptional efficacy in predicting noise annoyance,
achieving an impressive R-squared value of 0.99. Additionally, the Semantic Differential Method (SDM)
showcases its utility in predicting psychoacoustic parameters, with no R-squared value lower than 0.88.
Moreover, annoyance predictions calculated with spectrum-based parameters reported even more impressive results. However, challenges are encountered in reliably predicting sharpness, attributed 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. Compared to the work of my colleague Favaretto was introduced to deepen the validity of linear regression in the cab of the tractor, which brought valid results
and interest for research, confirming it a useful choice for research purposes. In addition, a further step was
introduced, free from bibliographical research, such as a methodology for translation of psycho-acoustic
parameters from English to Italian with the help of three native and bilingual subjects. This was done to have
a semantic scale and terms as clear and precise as possible, so that they can be submitted to an audience of
native Italian speakers without losing the semantic nuances of English terminology.
The aim of this research is to develop a study methodology which will improve the acoustic conditions in offroad vehicle cabins, thus improving the quality of work for farmers. Finally, the aim was to lay the foundations
for the search for an effective predictive model, based in future on neural networks, which would make it
possible to improve the quality of work in this category
Full paper
https://webthesis.biblio.polito.it/secure/33975/1/tesi.pdf
Keywords
tractor cabin acoustic comfort, HVAC noise impact, psychoacoustic analysis, subjective and objective noise evaluation, predictive noise modeling in agricultural vehicles