Análise em alta resolução de perfis magnéticos de sensores a laço indutivo para classificação de veículos automotores

• Ano de defesa: 2008
• Autor(a) principal: Hermes Aguiar Magalhaes
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais
• Palavras-chave em Português: Engenharia elétrica
• Link de acesso: https://hdl.handle.net/1843/BUOS-8CCF3L

Recent research efforts have been made to extract additional information from multiple vehicle detector technologies that, once processed in a more intelligent way, can lead to widearea/ section-related classification measurements, as opposed to limited point measurements, usually collected by traffic professionals. With efforts targeting vehicle classification, the current work presents a broad view of various current technologies used for classification, speed measurement and anonymous reidentification of vehicles through their paths. For reasons that span from technology robustness to low costs of implementation, the focus is on the magnetic profile analysis of inductive loop detectors (ILD). A proposition is made to use a high-resolution frequency estimation technique the Pisarenko Harmonic Decomposition PHD for vehicles magnetic profile extraction from signals acquired directly over inductive loops, excited by standard loop detectors commercially available. Following, a proposition is made to measure speed through use of cross-correlation between signature pairs in a speed trap loop configuration, when the attained accuracy improvement of speed calculations is investigated. Due to time multiplexing of loop channels performed by commercial detectors toavoid crosstalk, a modified cross-correlation estimator is implemented, capable of extracting the best possible estimation from non-uniformly sampled signals, without the need for interpolation. The aforementioned techniques are investigated through experimental laboratory essays over real metallic structures, when the contribution of eddy currents and ferromagnetic effect on magnetic profile shaping are qualitatively discussed. A prototype is constructed and used in experiments, capable of collecting in-the-field real vehicle data, including loop sensor signals and images of corresponding vehicles. Its software incorporates an innovative concept of service-oriented architecture: a generic delegation framework is built, targeting the development of multitasking real-time distributed system applications. This framework is then for the first time instantiated and validated, with target application being the software of the prototype. This architecture lets the programmer adjust the efficiency-traceability tradeoff in a multithreading environment, what is made possible through choice of in-depth level of operation for publishing services and for requesting services. The delegate mechanism embedded in this architecture also allows a potentiallyhigher efficiency in service dispatch with complex data to exchange, through on-demand specialization of dispatching class, since it is a template class typed by the exchanged data types themselves. Additionally, these data types are verified at compilation time, when theyare examined and used upon binding to the method that will be executed. In this way, programming errors can be caught earlier in software development cycle.