Modelo computacional para el análisis avanzado de señales de ruidos ambientales

Abstract

This thesis presents a computational model for advanced signal analysis of ambient noise from Mexico City. This model incorporate characterization of the equivalent continuous sound level (NSCE) recompiled by the Sistema de Monitoreo Ambiental de Ruido de la Ciudad de México (SIMAR-CH), the analysis of pattern per day, lapse, hour and every five minutes. For the characterization of patterns, the data are sampled and recompiled in ten points (nodes) distributed around Centro Histórico. This sampling is done every five minutes; signal is measured and it is sent to the system center, there, a weighting is realized for calculating NSCE. This information is saved in weighting every five minutes and each hour. For NSCEs characterization, data is extracted from SIMAR-CH, patterns are extracted per day, lapse and hour. The patterns per day are used for knowing certain irregularities for uses and customs of society; it means the behavior of sounds level depends of the social events in certain days. The lapse patterns determine the irregularities of noise in routine events. These events occur in certain lapses of day. The patterns per hour are analyzed for finding the NSCEs, which determines the irregularities. Apart from the analysis per day, lapse and hour, the NSCEs are analyzed every five minutes. The analysis of the data in every five minutes realizes into two phases: the first is used for knowing critical events during the hour. This analysis starts extracting the highest level, peaks and falls (levels which suffer an important changed on account of five minutes). The second analysis realizes directly from the signal. In this first analysis we look in a knowledge base of certain events to know the origin of the anomaly. Once the first analysis have done and finished the critical events, we make a spectral analysis of the determined noise signal. This Project aims to be the analysis base of the independent components of signals. This analysis realizes the component extraction and classification.