ECTI Transactions on Electrical Engineering, Electronics, and Communications

Hybrid Movie Recommendation System with Content-Based and Memory-Based Collaborative Filtering based on Deep Neural Network

2024-09-23T07:30:30+07:00

Recommendation systems assist users in filtering qualified data from vast datasets. Content-based and collaborative filtering techniques are the most widely used among the various recommendation models. In recent years, neural network algorithms have been employed to tackle recommendation problems. How- ever, while these methods reduce the error between true and predicted values, personalized recommenda- tions remain relatively lacking. To address this, we propose a movie recommendation system based on deep neural networks and user vocabulary preference features to alleviate cold start and sparsity issues, re- duce prediction errors, and improve recommendation efficiency. We evaluate our model using hit rate (HR) and average reciprocal hit rank (ARHR) as indica- tors, achieving an HR of 0.76 and an ARHR of 0.38. The robustness of our model is demonstrated through comparisons with other studies.

Optimal Coordinated Voltage Control of Distribution Networks Considering Renewable Energy Sources

2024-08-21T12:42:05+07:00

In light of the pressing concerns regarding global warming and the diminishing availability of fossil fuels, there has been a notable surge in the adoption of distributed generation (DG) systems, which harness clean and renewable energy sources. These systems are strategically placed in close proximity to end-users, thereby reducing power losses. However, without effective control mechanisms, issues such as voltage instability and increased power losses can impede the efficient functioning of the power grid. In this paper, an innovative approach termed Optimal Coordinated Voltage Control (OCVC) designed for distribution networks integrating dispersed renewable energy sources. Employing a genetic algorithm (GA) methodology and implementing multi-core simulation using the open-source platform OpenDSS, this method aims to optimize the settings of voltage control devices remotely. Moreover, it accounts for dynamic variations in both load and generation patterns through day-ahead scheduling. To evaluate the efficacy of the proposed technique, simulations are conducted on a test distribution network featuring the integration of DG systems.

Bi-level optimization algorithm for trading quantity and surplus maximization in P2P electricity market

2024-12-01T17:50:11+07:00

The increasing adoption of renewable energy and the evolution of energy markets have led to the need for innovative trading mechanisms, particularly in peer-to-peer (P2P) energy markets. This paper proposes a bi-level optimization algorithm for trading quantity and surplus maximization (BLO-TQSM) in P2P energy trading, incorporating a double-side carbon taxation scheme (DCTS). The BLO-TQSM algorithm is designed to optimize both the trading quantity and surplus by finding the best matching of participants in the market, while the DCTS mechanism integrates carbon tax considerations into the pricing of fossil and renewable energy sources. The shift factor, obtained by particle swarm optimization (PSO), is introduced to find the proposed bi-level maximization algorithm. The proposed method was tested in two scenarios: one without DCTS and one with DCTS. The results show that the algorithm significantly improves trading quantity and surplus in the P2P market compared to traditional power pool models. Moreover, the inclusion of DCTS further enhances the market's environmental sustainability by promoting the use of renewable energy and moving towards a carbon-neutral market.

Transformer Parameters and Transmission Line Temperatures Estimation in Three-phase Power Networks

2024-09-24T14:07:42+07:00

This paper presents a method for temperature-dependent three-phase state estimation where not only bus voltage phasors but also tap ratio, leakage admittance parameters of transformers, line conductor temperature, and weather environment variables are considered as the state variables. The three-phase tap-changing transformer model and the weather-based transmission line model are integrated into the proposed state estimation, which is then formulated as a nonlinear optimization problem based on the weighted least squares objective function. The proposed method is demonstrated on the IEEE 30-bus and 118-bus systems modified as three-phase test systems. Simulation results indicate that the proposed formulation outperforms the conventional state estimations, which do not consider transformer parameters and line temperature, in terms of estimation accuracy and bad data detection performance.

Comparison of Clearances Applying IEEE and IEC Standards for 230 kV Air Insulated Substation

2024-12-11T20:46:55+07:00

This paper analyzes the insulation levels and electrical air clearances of 230 kV air insulated substation (AIS). In order to get the basic lightning impulse insulation level (BIL) and clearances, the crest voltages at any equipment are required. By applying calculation method taken from IEEE Std. 1313.2, IEEE Std. 1427 and IEC 60071-2 the insulation levels and clearances have been found out and compared. The simplified calculation and simplified digital simulation via EMTP-ATPDraw have also been analyzed. The effect of elevation from sea level as well as the switching surge have been considered for insulation levels and clearances as recommended by the standards. According to IEEE Std. 1427 which taking into account the basic switching impulse insulation levels (BSL), the iterative method is inevitably required. Moreover, at this voltage level, the procedure for calculations refer to IEC 60071-2 in range I can also be applied. To compile with IEC 60071-2 for calculations the insulation levels and clearances, the iteration process accounting for BSL is not required. The clearances applying IEEE and IEC standards are closed and not significantly difference.

The Assessment of Mildly Aged XLPE Insulation using Polarization – Depolarization Current measurement

2024-10-09T14:46:05+07:00

Insulation systems used in the power system have to interact with the environment, which leads to its ageing and, eventually, failure. Cross – Linked Polyethylene (XLPE) insulation of underground cables is a vital component of the power system. Condition of XLPE underground cable insulation has a significant impact on the reliablity and economical operation of the electrical power grid. Several techniques have been in use for the condition assessment of underground cables. Polarisation and Depolarisation Current (PDC) measurement method has been commonly used to assess moisture and the ageing condition of the underground cable insulation system. In this work, dielectric response of XLPE samples, aged under different humidity and temperature conditions are investigated. The PDC measurement is carried out on unaged and aged samples. The XLPE samples are aged under constant stress and cyclic stress. The PDC measured data are used to estimate Degradation Index and Loss Factor. The depolarization characteristics obtained from PDC measurement, is further processed to obtain equivalent electrical circuit parameters i.e. resistance and capacitance of XLPE insulation based on Extended Debye Model. Using Model parameters, dissipation factor and ageing factor are estimated under different ageing conditions.

Ensemble Machine Learning for Identifying Fake News Headlines in Thailand

2024-09-06T07:25:00+07:00

Advancements in information technology (IT) have rapidly transformed news dissemination across on- line platforms, creating challenges in analyzing di- verse, often non-credible news sources. This issue is especially notable in Thai headlines due to linguis- tic nuances and broad topic range. Consequently, individuals are susceptible to the dissemination of false information, resulting in emotional, financial, and stability damages. To address this concern, this research proposes the utilization of term frequency- inverse document frequency (TF-IDF) in conjunc- tion with various machine learning (ML) models as a multi-classifier, i.e., multinomial naïve Bayes (MNB), k-nearest neighbor (KNN), support vector machine (SVM), and extreme gradient boosting (XGBoost) al- gorithms. All classifier models can classify fake news from the news headlines within the Thai context. To enhance the classification accuracy, ensemble ML techniques, such as fuzzy integral and blending, were applied. A dataset comprising news headlines from the Anti-Fake News Center Thailand, under the Min- istry of Digital Economy and Society, is constructed for analysis. The proposed ensemble blending tech- nique with logistic regression demonstrates a com- mendable accuracy rate of up to 97%, underscoring its efficacy in distinguishing authentic news from mis- information.

Inverse Design of 1×2 MMI Based on Automatic Differentiation in Silicon Photonics

2024-12-08T15:35:06+07:00

Silicon photonics has experienced significant advancements in recent years, driven by the increasing demand for efficient and high-performance optical components within integrated circuits. A crucial technique in this realm is the inverse design method, which is instrumental in optimizing photonic devices. Inverse design utilizes computational algorithms to identify the optimal configuration of a device based on specific performance criteria, making it a powerful tool for enhancing silicon photonics. The integration of Automatic differentiation into the inverse design process has further revolutionized this approach by improving the precision and efficiency of optimization. This technique enhances the ability to fine-tune design parameters and achieve desired device characteristics. The 1x2 Multi-Mode Interferometer (MMI) plays a vital role in optical functions such as signal splitting, combining, and routing. Its significance in various photonic applications underscores the importance of precise design and optimization. Therefore, in this study, we focus on applying automatic differentiation (AD) to optimize the inverse design of a 1x2 MMI with a function of 3dB splitting power, aiming for minimal size to facilitate easy integration into optical systems

West First Turn Routing Algorithm with Backtracking Mechanism for on Chip Networks

2024-07-24T10:45:41+07:00

In order to design high-performance and energy-efficient electronic systems, this research article is an attempt to investigate the usage of Network-on-Chip architectures as a possible substitute for conventional bus-based interconnects in the integrated circuits. This study primarily focuses on backtracking strategies for assessing the West First Turn routing protocol. Increasing the scalability and efficiency of Network on Chip based electronic systems by analyzing parameters like energy usage and resource utilization. This article presents a comparative study of XY routing, West First Turn routing, and Odd-Even Turn routing along with the West First Turn routing with backtracking mechanism for handling such issues including wormhole flow control in the Network on Chip architecture. Simulations results for the for a 3x3 mesh network coded in Verilog-HDL using Xilinx Vivado 2021.1 infers that West First Turn routing with backtracking mechanism is found better in terms of simplicity, adaptability, and resource utilization efficiency. This article will help researchers in proffering valuable insights and inferences for their research and development work as a contribution to the comprehension of associated challenges, potential solutions, and advancements for the selecting an appropriate routing mechanism in accordance with particular design specifications. The knowledge shared in this study not only advances Network on Chip technology but also offers useful recommendations for creating scalable and effective computing systems.

Power Estimation in Three-Phase PWM Rectifiers Using ANN

2024-09-23T15:45:17+07:00

At the industrial consumer level, an unbalanced threephase supply is very common, which creates control issues with PWM rectifiers. Direct power control (DPC) has been a proven controller in the field of rectifier control. The DPC requires the estimation of active and reactive as these quantities are used as the control variable in the loop. For an unbalanced grid, the
power estimation becomes complex with the extended pq theory. This article presents a novel approach to estimating the instantaneous power using an artificial neural network (ANN), streamlining the complex control structures involving back-to-back connected secondorder generalized integrator (SOGI) blocks. The intricate interconnection of multiple SOGI blocks poses challenges in terms of computational complexity. The proposed method leverages ANN to replace the SOGI blocks, which makes the system simpler and more efficient. Rigorous
simulation using MATLAB Simulink and comparative studies reveal that with ANN, the performance of the based system can be replicated, and a reduced computational burden can be achieved, leading to improved realtime response. Real-time simulation in RT-LAB validates the proposed ANN-based method for PWM rectifiers’ power estimation. Integrating ANN models ensures accurate emulation of complex control structures, affirming the approach’s efficacy for industrial deployment.

A Combination Approach for 2-D Interference Channel Based on Factor Graph in BPMR System

2024-06-18T16:59:24+07:00

High areal density in bit patterned media recording (BPMR) system encounters several challenges such as two-dimensional (2-D) interference channel and media noises in the presence of severe fluctuations. To combat the effects of the 2-D interference channel, a factor graph-based detector (FGB) is one of the promising detectors for the 2-D detection which can mitigate the channel impairment. In this paper, we propose an improvement of factor graph decoding for 2-D interference channels of BPMR system. The 2-D interference channel is separated into the two targets that are an upper triangular target and the lower triangular target. After that, the joint outputs are averaged of each detector. The simulation results show that the proposed FGB detector method achieves lower the bit error rate (BER) performances than both the conventional FGB detector and 2-D FGB detector on BPMR system with an areal density of 3 and 4 Tb/in².

Implementing a Triple L-shape Resonator on a Monopole Antenna for Multiband Operation in WLAN, WiMAX, 4G LTE, and 5G Sub-6G Networks

2024-12-13T07:20:06+07:00

The present research proposes an innovative multiband antenna design including a triple L-shape structure. This design is specifically targeted for utilization in a wireless local area network (WLAN), worldwide interoperability for microwave access (WiMAX), 4G LTE, and 5G Sub-6G networks. The antenna being considered is designed in an L-shaped layout, consisting of three separate components. The triple L-shape demonstrates the capacity to independently produce resonant frequencies. The middle L-shape, which is the longest, is notably responsible for generating the initial resonance frequency at 1.8 GHz. In addition, the L-shaped structure on the right exhibits a secondary resonance frequency of 2.45 GHz, whilst the L-shaped structure on the left generates a tertiary resonance frequency of 3.6 GHz. The proposed antenna demonstrates an extra resonance frequency of 4.6 GHz, which fits within the application frequency range of 5.2 GHz. The presence of the L-shape structure in the middle and right side of the suggested antenna arrangement is responsible for generating the 4^th resonance frequency, which can be attributed to the harmonic frequency. The antenna being discussed is fabricated on a FR4 substrate, measuring 30×50 mm² in size. After performing testing on the antenna, it is evident that both the simulation and measurement results demonstrate a suitable response across the whole operational frequency range. The antenna is capable of operating efficiently at specific frequencies. These frequencies include 1.8 GHz (1.71 GHz - 1.89 GHz) for LTE 1800, 2.45 GHz (2.39 GHz - 2.71 GHz) for IEEE 802.11b&g WLAN systems, LTE 2600, and 5G Sub-6G network, and 3.6 GHz (3.2 GHz - 5.37 GHz) for 5G Sub-6G network, WiMAX system, and IEEE 802.11a WLAN system. The success of this performance is demonstrated by the magnitude of the reflection coefficient, represented as |S₁₁|, which consistently stays below -10 dB. The antenna's average gain across all operational frequencies is approximately 2.5 dBi. Moreover, the antenna's radiation pattern maintains a consistent omnidirectional characteristic over the frequency bands of 1.8 GHz, 2.45 GHz, and 3.6 GHz. However, the radiation pattern at a frequency of 5.2 GHz displays distortion due to the antenna's higher-order mode. Furthermore, the antenna maintains its efficacy in wireless communication systems, such as devices that combine Wi-Fi and mobile cellular 4G&5G technologies in a single unit, occasionally called pocket Wi-Fi + 4G&5G cellular.