مدلسازی ضریب جذب صدا برای پانل‌های میکرومتخلخل طبیعی چوب کاج یک لایه و دو لایه

Microporous absorbers (MPP) as green absorbers are widely used in acoustic design and noise control. The purpose of the present study is to model the sound absorption coefficient in natural microporous panels of single-layer and double-layer pine wood.
Methodology: In this research, the experiment was designed based on full factor. Single-layer and double-layer microporous panels were made of pine wood. Next, the effect of using composites made of pine wood fibers and shavings with densities of 200 to 350 kg/m3 and variable thickness of 0.5 to 2.5 cm on the sound absorption of microporous panels made in the previous stages was investigated. To measure the sound absorption coefficient, the tube impedance device made according to the ISO10534-2 standard was used, and non-parametric statistical tests such as Kruskal-Wallis and Mann-Whitney were used. All analyzes were analyzed by SPSS statistical software version IBM SPSS STATISTIC 27. Algorithms based on artificial neural networks, support vector machine and machine learning were used for modeling.
Results: The results showed that double-layer panels have a higher sound absorption coefficient than single-layer panels. Kruskal-Wallis test to check the mean and average absorption coefficient in three different thicknesses was significant (pvalue<0.05). In single-layer panels, changing the air gap caused a significant increase in the sound absorption coefficient (p-value=0.006<0.05). In double-layer panels, changing the air gap caused a significant change in the average absorption coefficient of the panels. The result of the test in double-layer panels is significant (pvalue<0.05), so the null hypothesis is rejected and the opposite hypothesis is accepted in the case of these woods.
Conclusion: The use of machine learning models compared to the regression model causes a more accurate estimation of the absorption coefficient of the panel. Also, the thickness of the panel has an effect on the absorption coefficient. It is possible to predict the absorption coefficient by selecting the thickness, porosity diameter and number of layers and other characteristics as well as choosing the right wood. In general, considering the error rate of machine learning models, it can be said that these models have a better ability to estimate the sound absorption coefficient and estimate the absorption coefficient with less error than the regression model.