Abstract-Finite States Model Predictive Control (FS-MPC) appears as a promising control technique to be applied to power converters in the industry. However, the FS-MPC presents some drawbacks as non constant switching frequency and high sampling frequency. This work proposes a FS-MPC with constant switching frequency and low sampling frequency applying a Discrete Space Vector Modulation (DSVM) technique. The real state vectors of the converter are used together with new virtual state vectors forming switching sequences each sampling period. The advantages and disadvantages of the proposed FS-MPC with the DSVM are analyzed using a two-level three-phase inverter connected to the grid as setup to introduce the proposed technique. Simulation results are presented, showing that using the proposed technique the switching frequency is fixed and the sampling frequency can be lowered without reducing the quality of the converter behavior.
Resumen:Este artículo presenta el diseño de un controlador robusto H∞ usando técnicas de desigualdades matriciales lineales (LMI), para controlar la posición de Pitch y de Yaw en un helicóptero. Se presenta el diseño de un controlador FF+LMI con el propósito de conseguir la estabilización del sistema, y adicionalmente se realiza el diseño de un controlador FF+LMI+Integrador, para hacer que el error de seguimiento sea igual a cero. Posteriormente se presentan los resultados de las simulaciones sobre el modelo no lineal del sistema, así como una comparación con los controladores FF+LQR y FF+LQR+I desarrollados por los fabricantes del dispositivo. Abstract:This article presents the design of a robust H∞ controller using techniques of linear matrix inequalities (LMIs) for controlling the position of Pitch and Yaw in a helicopter. Designing a FF + LMI controller is presented for stabilizing the system, and further the design of LMI + FF + integrator controller is performed to make the tracking error equal to zero. Thereafter the results of simulations are showed on the nonlinear system model and a comparison with FF+LQR and FF+LQR+I which were developed by device manufacturers.Palabras clave: control robusto, desigualdades matr. iciales lineales, modelo nominal, sistema no lineal.
Co-diseño de una aplicación para el reconocimiento in situ del gorgojo de los andes en cultivos de papa Resumen -Diversos métodos son empleados evitar que los cultivos de papa se vean afectados por enfermedades y plagas, uno de estos, es el monitoreo que consiste en personas encargadas de recorrer los cultivos y emplear sus capacidades cognitivas para reconocer la presencia de plagas. Sin embargo, limitaciones en la capacidad humana como la imprecisión por la subjetividad introducida por el agricultor pueden ocasionar fallas en el diagnostico. Por esta razón, se implementó un sistema capaz de detectar la presencia del gorgojo de los Andes. Para esto se emplea visión artificial para realizar el preprocesamiento de imágenes extraídas de fotografías proporcionadas por agricultores. Además, se desarrolla un modelo de aprendizaje profundo basado en la arquitectura VGGNet. La arquitectura fue llevada a una aplicación móvil mediante la herramienta el modelo denominado MobileNet. Los resultados arrojaron, un índice de reconocimiento adecuado, obteniendo una precisión en la predicción hasta del 84%. Palabras claves --Aprendizaje profundo, red neuronal, foliolo, preprocesamiento, aplicación móvil Abstract -Various methods are employed to prevent potato crops from being affected by diseases and pests, one of which is monitoring, which consists of people walking through the crops and using their cognitive abilities to recognize the presence of pests. However, limitations in human capacity such as inaccuracy due to the subjectivity introduced by the farmer can cause failures in the diagnosis.For this reason, a system capable of detecting the presence of the Andean weevil was implemented. For this purpose, artificial vision is used to perform the preprocessing of images extracted from photographs provided by farmers.In addition, a deep learning model based on the VGGNet architecture was developed. The architecture was taken to a mobile application using the model called MobileNet. The results showed an adequate recognition rate, obtaining a prediction accuracy of up to 84%.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.