Student EEICT 2021

Browse

Recent Submissions

Now showing 1 - 5 of 128
  • Thumbnail Image
    Item
    Smart Home Wired Architecture
    (Vysoké učení technické v Brně, Fakulta elektrotechniky a komunikačních technologií, 2021) Sykora, Tomas
    This article deals with the problem of Smart home systems. Electrical installations innew built and older buildings have recently integrated smart devices. These devices use IoT technology,which needs to communicate with each other. This article discusses the possibilities of replacingmost often used wireless communication for a new type of wired based on Industry IoTstandards. Using wired technologies will enhance the system reliability, safety and security. Whatmore it involve reduction of the electromagnetic smog, a potentially health-endangering effect generatedby devices communicating wirelessly at high frequencies.
  • Thumbnail Image
    Item
    Techniques For Avoiding Model Overfitting On Small Dataset
    (Vysoké učení technické v Brně, Fakulta elektrotechniky a komunikačních technologií, 2021) Kratochvila, Lukas
    Building a deep learning model based on small dataset is difficult, even impossible. Toavoiding overfitting, we must constrain model, which we train. Techniques as data augmentation,regularization or data normalization could be crucial. We have created a benchmark with a simpleCNN image classifier in order to find the best techniques. As a result, we compare different types ofdata augmentation and weights regularization and data normalization on a small dataset.
  • Thumbnail Image
    Item
    Influence Of Piezoelectric Material Properties On Impedance Characteristic And Improvements Of Calibration Equipment
    (Vysoké učení technické v Brně, Fakulta elektrotechniky a komunikačních technologií, 2021) Krejčí, Jakub
    This paper deals with finite element simulations of impedance characteristics piezoelecticelements utilized in sensors of acoustic emission (AE). Coefficients of piezoelectric material arechanged to evaluate the sensitivity to their variation in case they are not known precisely, so to whichextent can they influence the sensitivity of the sensor. Result showed sensitivity on coefficient inc-matrix, mainly to coefficients c11 and c13. Second investigated topic is issue with test stand for calibrationof AE sensors, where output of force sensor was affected by parasite signal. By contactlessmeasurement with laser interferometer was discovered the parasite signal is caused by mechanic ofthe system and it is no threat for the measuring chain.
  • Thumbnail Image
    Item
    In-Bed Posture Classification
    (Vysoké učení technické v Brně, Fakulta elektrotechniky a komunikačních technologií, 2021) Husák, Michal
    The growing trend of the population age contributes to the accumulation of patients insocial facilities and in-home care, which leads to growing chronic diseases. Modern systems try toimprove the effectiveness of health care interventions. Our work aims to create a widely applicableplatform that combines the measurement of in-bed position with another’s negative states. All thesephysical influences are mainly the cause of chronic tissue damage (pressure ulcers). Processing ofthe pressure distribution on the bed is a more dimension problem. The mentioned data are multimodal.Therefore, we used the machine learning (ML) method to obtain the properties.
  • Thumbnail Image
    Item
    Derivation And Practical Comparison Of Recursive Ls And Tls System Identification Methods
    (Vysoké učení technické v Brně, Fakulta elektrotechniky a komunikačních technologií, 2021) Friml, Dominik
    The least squares (LS) type of methods are the most widely used methods in system identificationdespite their obvious imperfection. Such methods use a regressor, that is supposed not tocontain any error, notwithstanding that it is constructed from measured data. This can be solved byusing the total least squares (TLS) type of methods. Derivation of both batch and recursive methodsof LS and TLS for identification and their practical comparison is presented in this paper