Valoración de la molestia percibida por el ruido de tráfico aplicando redes neuronales artificiales
This work covers the configuration and application of an artificial neural network, to make an assessment of perceived verbal annoyance between people towards different descriptors of traffic noises such as A-weighted equivalent sound pressure level (LA), equivalent continuous sound pressure level (...
Gorde:
| Egile nagusia: | |
|---|---|
| Formatua: | bachelorThesis |
| Hizkuntza: | spa |
| Argitaratua: |
2017
|
| Gaiak: | |
| Sarrera elektronikoa: | http://dspace.udla.edu.ec/handle/33000/7437 |
| Etiketak: |
Etiketa erantsi
Etiketarik gabe, Izan zaitez lehena erregistro honi etiketa jartzen!
|
Etiketa erantsi | Gaia: | This work covers the configuration and application of an artificial neural network, to make an assessment of perceived verbal annoyance between people towards different descriptors of traffic noises such as A-weighted equivalent sound pressure level (LA), equivalent continuous sound pressure level (LAeq), percentile level (Ln), temporal sound level variance (TSVL), peak factor (CF) and spectral centroid (G). All these values were obtained in a data survey carried out in a previous work, which presents responses of verbal annoyance tabulated through surveys. The configuration of the neural networks was made using MatLab software, where the architecture of the network is specified based on parameters such as number of neurons, layers and the error to be taken into account to verify its operation, which in this work falls into the mean square error. Each parameter of the network architecture was modified several times obtaining different results. These results were compared to each other until finding a configuration that guarantees a minimum mean square error and therefore a correct functioning. Finally, a comparison was made between the final data delivered by the neural network and those presented in the previous work in order to find the efficiency in the neural network. |
|---|