Detección secuencial de vibraciones discriminables por rangos espectrales en placas delgadas.
Archivos
Autores
Castellanos Domínguez, César Germán
Director
Tipo de contenido
Artículo de revista
Idioma del documento
EspañolFecha de publicación
2001-10
Título de la revista
ISSN de la revista
Título del volumen
Documentos PDF
Resumen
Se presenta un método de detección de vibraciones que posean cualidades espectrales claramente diferenciables, a priori conocidas a la salida de una placa resonante delgada. Se desarrolla la síntesis del detector secuencial en tiempo continuo, basados en el criterio de Neyman-Pearson. El algoritmo está orientado a tareas en las cuales se exige tomar la decisión sobre la presencia de un posible modo de vibración a la entrada del detector, que sea diferenciable sobre rangos de frecuencia espectralmente separables. El detector secuencial simplificado para un solo rango de frecuencia diferenciable y una sola clase de OD a distinguir fue realizado y probado experimentalmente, demostrando un desempeño aceptable y reducidos costos de implementación.
Abstrct : A method is presented for detecting of vibration signals having priori well-known frequency qualities of discrimination. Signals are taken at the output of the thin resonant membrane. The sequential detector synthesis is carried out for continuous time, based on the Neyman-Pearson approach. The algorithm allows to make the decision on the presence of only one vibration mode at the detector input among a full set of different modes, each of them being spectrally separable. Although, a general expression for a finite number of classes is obtained, this paper develops a simplified sequential detector for a single class to be distinguished over a background noise. Experimental testing was accomplished for a case of detection of fruits with two different densities, and it showed a good trade-off between performance and cost implementation.
Abstrct : A method is presented for detecting of vibration signals having priori well-known frequency qualities of discrimination. Signals are taken at the output of the thin resonant membrane. The sequential detector synthesis is carried out for continuous time, based on the Neyman-Pearson approach. The algorithm allows to make the decision on the presence of only one vibration mode at the detector input among a full set of different modes, each of them being spectrally separable. Although, a general expression for a finite number of classes is obtained, this paper develops a simplified sequential detector for a single class to be distinguished over a background noise. Experimental testing was accomplished for a case of detection of fruits with two different densities, and it showed a good trade-off between performance and cost implementation.