Naruttam Kumar Roy | Searcharxiv

arXiv2025

Impact Analysis of COVID-19 in Bangladesh Power Sector and Recommendations based on Practical Data and Machine Learning Approach

This paper investigates the impact of COVID-19 on the power sector in Bangladesh, how the country has dealt with it, and explores the path to stability. The study employs data visualisation and complex statistics to examine critical data about power systems in Bangladesh. This includes load patterns on a daily, monthly, annual, weekend, and weekday basis. Significant alterations in these patterns have been observed during our study e.g., in April and May of 2020, the power demand decreased by approximately 15.4% and 17.2%, respectively, compared to the corresponding period in 2019. We have used a Long-Short-Term Memory (LSTM) framework to predict the load profile of 2020 excluding COVID-19 effects. This model is compared with the actual load profile to determine the degree to which COVID-19 has impacted. The comparison indicates that the average power demand decreased by approximately 19.5% in April 2020 and 18.3% in May 2020, relative to its projected value. The study also investigates system stability by analyzing transmission loss and load factor, and the environmental effect by analyzing the Carbon Dioxide emission rate. Finally, the study provides recommendations for overcoming future disasters, such as developing more resilient power systems, investing in renewable energy, and improving energy efficiency.

eess.SY↗

arXiv2024

Review of meta-heuristic optimization algorithms to tune the PID controller parameters for automatic voltage regulator

A Proportional- Integral- Derivative (PID) controller is required to bring a system back to the stable operating region as soon as possible following a disturbance or discrepancy. For successful operation of the PID controller, it is necessary to design the controller parameters in a manner that will render low optimization complexity, less memory for operation, fast convergence, and should be able to operate dynamically. Recent investigations have postulated many different meta-heuristic algorithms for efficient tuning of PID controller under various system milieus. However, researchers have seldom compared their custom made objective functions with previous investigations while proposing new algorithmic methods. This paper focuses on a detailed study on the research progress, deficiency, accomplishment and future scopes of recently proposed heuristic algorithms to designing and tuning the PID controller parameters for an automatic voltage regulator (AVR) system. Objective functions, including ITSE, ITAE, IAE, ISE, and ZLG, are considered to enumerate a measurable outcome of the algorithms. Considering a slight variation in the sytem gain parameters of the algorithms, the observed PID gain with ITSE results in 0.81918 - 1.9499 for K_p, 0.24366 - 1.4608 for K_i, and 0.31840 - 0.9683 for K_d. Whereas with ITAE the values are 0.24420 - 1.2771, 0.14230 - 0.8471, and 0.04270 - 0.4775, respectively. The time domain and frequency domain characteristics also changes significantly with each objective function. Our outlined comparison will be a guideline for investigating any newer algorithms in the field.

eess.SY↗