Sistema neural antifurto veicular

• Autor(a) principal: Ramos, Celso de Ávila
• Data de Publicação: 2016
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFLA
• Texto Completo: http://repositorio.ufla.br/jspui/handle/1/29916
• topic: Redes neurais artificiais; Antifurto veicular; Classificação de condutores; Segurança veicular; Artificial neural networks; Vehicular anti-theft; Conductors classification; Vehicular safety; Ciência da Computação

Currently, the concern for the safety of properties has constantly been among the population, especially in countries where the theft rates are high. Faced with a worrying scenario, issues about developing technologies and solutions that are able to reduce theft rates must be addressed, seeking to improve existing techniques and / or develop new ones. This study aims to verify the viability of using artificial neural networks for the detection of unauthorized driving of vehicles and implement an automated real-time system, based on an artificial neural network trained to classify the driver as to how they drive the vehicle, based on data obtained from the automobile itself. Therefore, we used the OBD-II device commonly used to obtain data from vehicle sensors. Variables like throttle position, acceleration in x, acceleration in y and acceleration in z were used as inputs to a neural network to classify the driver either as authorized or not authorized to drive the vehicle. An Android app that sends data from the OBD-II to a Web Service Python was developed. This Web Service has a scan function that uses a neural network trained to classify the driver and return an answer to the user. The training algorithm used was backpropagation, obtaining satisfactory results during the tests, with 88% of the trained neural network hits. The test of the efficiency ratio was measured by the Kappa coefficient, with a result as excellent for this index. The Neural Vehicle Anti-Theft System is a tool that can help owners monitor the driving of their car. It is hoped, too, that the system can help other areas of interest, as authorities and insurance companies. The use of Artificial Neural Networks to classify the driver was proved to be feasible and effective for this purpose. It is also important to note that the OBD-II device can be used for other purposes that go beyond the diagnosis of vehicle components for its proper maintenance. The developed system proved that it is possible to assess the behavior of the driver by means of data supplied by the vehicle they conduct.