Sound Quality–Based Acoustic Optimisation for Construction Machine Operators |

Sound Quality–Based Acoustic Optimisation for Construction Machine Operators

Eleonora Carletti and Francesca Pedrielli (IMAMOTER Institute)

As in many other fields of application, besides the mandatory provisions, the construction machine industry is now oriented towards the sound quality approach. Hence, at least in the last decade, research has been dealing with the identification of a set of acoustic and psychoacoustic metrics able to describe people’s auditory perception of noise signals with respect to the annoyance sensation.









The optimization process was applied to noise signals binaurally recorded at the operator position of two compact loaders with different dimensions and mechanical power. The recordings had a duration of 7-8 seconds and included both the loading phase and the movement of the machine. As in the case of stationary signals, the target of this optimization process was the simultaneous minimization of the objective parameters best related to the annoyance sensation. The numerical analyses were performed using the multiobjective genetic algorithm (MOGA) governed by the  modeFRONTIER optimization procedure. The calculation of loudness and sharpness values was therefore performed by a MATLAB script developed for this purpose. This script read the values of the time-frequency matrix as input and gave the array of the values of loudness and sharpness as output. The loudness values were calculated according to the procedure described in the DIN 45631/A1 standard in order to take into account the time variability of these noise signals. The output of the optimization process included a set of several solutions (Pareto Frontier). Each solution consisted of twenty seven dB-values representing the level variations, frequency by frequency, suggested by the optimization algorithm in order to minimize the objectives. The best solutions with respect to the minimization of sharpness (no.13 for signal A and no.15 for signal B) suggest significant noise reductions at medium-high frequencies, but, unfortunately, an increase of noise levels was found at low frequencies (40-630 Hz) for both signals. The best solutions with respect to the minimization of loudness (no.1 for both signals) suggest significant noise reductions all through the frequency range. Looking for a compromise between sound quality improvement and practical constraints, the right approach could be to start from the solution with the minimum value of loudness and proceeding to solutions with progressively higher loudness values until a feasible solution is found.