| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 1.69 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
O principal objetivo desta dissertação é apresentar um sistema de deteção de ignições florestais o qual é baseado num algoritmo de inteligência artificial utilizando a metodologia Fuzzy Logic. Este sistema vai ter a capacidade de identificar pequenos focos de incêndio através da análise do desvio dos dados lidos, dando assim um alerta, recorrendo para tal a sensores de chama, humidade e temperatura, localizados nas florestas em estudo. Este método consiste então em técnicas de aprendizagem supervisionada e não supervisionada, para fornecer uma série temporal de valores de perigo e que depende apenas de variáveis climáticas e sensores de chama como dito anteriormente. Posto isto a resposta de alarme será fornecida consoante o erro obtido através da comparação de uma base de dados com uma base de teste com valores lidos em tempo real na floresta.
The main objective of this dissertation is to present a forest ignition detection system which is based on an artificial intelligence algorithm using the Fuzzy Logic methodology. This system will have the ability to identify small fires by analyzing the deviation of the data read, this giving an alert, using flame, humidity and temperature sensors located in the forests under study. This method then consists of supervised and unsupervised learning methods, to provide a time series of hazard values and which depends only on climatic variables and flame sensors as stated above. That said, the alarm response will be provided depending on the error obtained by comparing a database with a test base with values read in real time in the forest
The main objective of this dissertation is to present a forest ignition detection system which is based on an artificial intelligence algorithm using the Fuzzy Logic methodology. This system will have the ability to identify small fires by analyzing the deviation of the data read, this giving an alert, using flame, humidity and temperature sensors located in the forests under study. This method then consists of supervised and unsupervised learning methods, to provide a time series of hazard values and which depends only on climatic variables and flame sensors as stated above. That said, the alarm response will be provided depending on the error obtained by comparing a database with a test base with values read in real time in the forest
Description
Keywords
Fuzzy logic Machine learning Incêndio florestal Sensores