ABSTRACT. Smart devices have evolved through voice control. They are permanently connected to the internet to provide information interactively. The electronic services offered are based on actions towards devices interconnected through wireless communication channels. They can be smart plugs, smart bulbs or any electronic device compatible with the smart audio system. Remote control via the Internet of these interconnected devices is possible. The advantages of these technologies are certain for various monitoring and services. The system proposed in this paper is able to monitor the predefined operation with online remote control. Especially for off-line situations, it intervenes through a local strategy to ensure a predictable functionality. Included here are emergencies, when power consumers are connected or there is critical assistance for certain remote services.

ABSTRACT. In this paper we develop the bond-graph modelling of sub-models using LC oscillators as parts of different detectors (pressure, metals, noise). These sub-models are realised using bond-graphs and are simulated using 20Sim software. For the study of the elements of magnetic circuits with localised parameters we used effort-flux analogies. In order to eliminate any type of divergence that could appear in the simulation of a non-linear circuit, the extended circuit will be restrained to a circuit where the proposed models for transistors (NPN and PNP) will be incorporated in sub-models. We used the detailed bond-graph models of the NPN and PNP transistors and we developed the bond-graph-model of the LC analysed sub-model.

ABSTRACT. In this paper we develop the bond-graph modelling of sub-models using sensors as parts of different detectors (temperature, motion, vibrations). These sub-models are modelled using bond-graphs and are simulated using 20Sim software. The advantages of bond-graph modelling of these sensors using 20Sim consist in the possibility to use sub-models used in different complex detectors and these sub-models of different nature could be modelled/integrated in a single (bond-graph) model and their behaviour can be analysed using 20Sim. The sensors are used to detect the state of a system and the information from the sensors is feeding the decision block/system. A sensor that measures the system response can affect or influence the behaviour of the system. The influence of such a sensor could be quantified studying the behaviour of the system and this behaviour can be described and studied using 20Sim, independently on the nature of the measure that influence the system. For example, the change of capacity could be used to measure forces, humidity, temperature and other kind of physical measures. All these parameters can be modelled and studied using a single and unitary model (bond-graph) and the simulation software 20Sim.

ABSTRACT. The paper analyses the influence of various parameters that influence the detection and recognition of traffic signs, to help people with visual impairments. We used the Raspberry PI 4 model B + platform and functions from the OpenCV library to reach our objective. 3 template matching functions from OpenCV were compared and we used the Normalized Cross Correlation (NCC) and Sum of Squared Differences (SSD) as matching metrics from the class of intensity based methods for template matching.

For the chosen methods (cross-correlation, inter-correlation and normalized square difference) the influence of the lighting on the traffic signs was analysed. For this, the histogram equalization method was applied to the images acquired during the evening, in order to improve the recognition of the traffic signs and the results were compared with the method without histogram equalization. Another parameter that influences the detection is the resolution of the template. All the analysed images had the same resolution (1440 x 1920), and the low-resolution templates (50 x 50 for example) did not give the expected results (the surface of the template represents 0.1% of the image to be analysed). The behaviour of the 3 methods was analysed and compared. Another source of errors is the shape of the template, the most suitable for detection being the rectangular one, the round ones being more often undetected or had false detection areas in the image. For the primary simulation, a program identical to the one on the Raspberry PI board was used, that was implemented on a PC. The final simulation was made on the Raspberry PI board, registering the computation times as well as the values of the local maximums (or minimums) for each case, in order to be able to analyse these data in the future by using detection thresholds. The results are presented in the form of tables, for a quick analysis, then the conclusions regarding the influence of the mentioned parameters on the detection of traffic signs are presented.

ABSTRACT. Exploring an unknown space it is one of the most well-known tasks to be performed by mobile robots. Among the most used sensors for obstacle detection there are ultrasonic sensors that have the advantage of simplicity in implementation as well as the possibility of obtaining a fast response time of the robot's control system.

ABSTRACT. This paper presents a comparative study between Turbo codes and Polar codes used for channel coding in mobile communications, and provide a description of these codes and their main characteristics. The performances of these codes are very close to Shannon capacity performance. The efficiency of Turbo and Polar codes is evaluated in term of bit error rate (BER) for a given value of signal to noise ratio (SNR) over Additive White Gaussian Channels (AWGN) by employing Binary Phase Shift Keying (BPSK) modulation scheme.

ABSTRACT. In industry, a step following the large-scale production is the high precision and high-speed counting of the parts to be delivered to the customer. The paper proposes an automatic system for counting parts. This is implemented with the help of a photoelectric emitting sensor, which sends an infrared light beam to a receiving photoelectric sensor, and when a piece on the moving conveyor belt interrupts the infrared light beam, the counter detects it. The electrical signal provided by the receiving photoelectric sensor, obtained because of detecting the presence of a piece, is applied to an electronic counter that increments (or decrements) the value in its memory. This automatic part counting system is reliable only if the pieces are opaque to the infrared light beam. If the parts have different holes, or are notched, the infrared light beam emitted by the transmitting photoelectric sensor to the receiving photoelectric sensor is interrupted several times for a single piece, and the counter records more than one unit. This is not desirable because it means that the pieces were counted incorrectly. The paper contain a figure showing some types of metal pieces manufactured for the automotive industry that can produce such counting errors. As we shall see the usual automatic counting system cannot be used to count the metal pieces if they have different holes, bevels, and notches. To create an automatic system capable of counting metal pieces with atypical shapes, we made use of the law of electromagnetic induction, and we built a customized sensor based on Eddy currents. The proposed method is applicable only to materials with good conductive qualities (metals, alloys, materials containing layers of conductive material). According to Lenz's law, an Eddy current creates a magnetic field that opposes the original magnetic field and so the Eddy currents return to the source of the magnetic field. Thus, the effect of eddy currents is the decreasing of inductance of the excitatory coil because the direction of the current induced in the material opposes the variation of the inductive current, decreasing the variation of the magnetic flux through the coil. The effect is visible near the coil of diamagnetic or weakly paramagnetic metals. If we get closer the coil to a ferromagnetic metal such as iron and nickel, this effect is masked by the increase in inductance due to the increase in the magnetic permeability of the environment. If the depth of penetration is greater than the thickness of the material then the device can be used to measure the thickness of the metal layer, otherwise it can be used to assess the diameter of conductors placed on the coil axis or the distance between the coil and the metal body (metal detector).

ABSTRACT. In this paper is presented a step-up five level inverter with reduced number of components. Only one dc voltage source, six active switches, two diodes and one capacitor are required. The problems of voltage balancing and numerous active switches are avoided. Two-pulse width modulation (PWM) control methods are investigated for the proposed inverter: phase disposition PWM (PD-PWM) and phase shift PWM (PS-PWM). For the two methods it has been studied how the number of the carrier signals affects total harmonic distortion factor (THD) and the harmonic distribution.

ABSTRACT. This article presents an incipient case study of learning music using Scratch programming environment and the impact of the theoretical musical knowledge on understanding and learning computer programming. Due to the fact that computer programming takes part of all major domains including soft skills, artistic and musical ones, it allows children to develop not only their technical skills but also their artistic and creative skills. In the same time, musical knowledge can be used as an amplifier in understanding computer programming which facilitates learning. The purpose of this article is to emphasize the development of children’s musical skills increased by their programming ones and the impact of theoretical musical knowledge on understanding and learning computer programming.

ABSTRACT. Artificial Neural Networks (ANN) are modelled on the workings of the human brain where information is transported through the synapses formed between individual neurons. The activation function decides the signal to be transmitted from a node of the ANN to the others that are connected, so we can interpret the value of the function as the "activation" of a neuron. There are many activation functions available, each with its advantages and disadvantages when it comes to learning rate and computational load. The current article presents the most commonly used activation functions and their applications in ANNs. We study the effect of using different activation functions for a convolutional neural network used for image classification. Model performance for the train and test set are presented for each fold of the k-fold cross-validation, as well as accuracy scores for the trained network.

ABSTRACT. In this paper are presented the results of the analysis of a series resonant inverter with phase-shift control. High signal and low signal models are obtained using the harmonic approximation method and contain both the power part and the frequency control loops, respectively the inverter power. The analysis of the small signal model is performed for various operating modes. Influences of changes in load parameters during heating on inverter stability have been studied.

ABSTRACT. This paper presents results that validate a complete mathematical model, obtained by harmonic approximation, of a series resonant inverter with phase control. Results obtained by simulation and confirmed by experimental determinations appear in parallel. Waveforms and measurement results confirm the analysis performed regarding the ZVS switching mode and also the amplitude-frequency characteristics of the frequency and power control loops for different values of the control angle and for if the load resistance changes are compared with those obtained by simulation