International Journal of Electrical, Energy and Power System Engineering

Digital Twin-Enabled Deep Reinforcement Learning for PV-ESS Energy Management in Active Distribution Networks

PhongMinh le — Tue, 30 Jun 2026 06:30:29 +0700

This paper proposes a real-time energy management framework for photovoltaic (PV) and energy storage system (ESS) integrated distribution networks using deep reinforcement learning (DRL) and Digital Twin (DT) technology. The objective is to improve operational efficiency under uncertain conditions while maintaining voltage stability and reducing battery degradation. The proposed method is validated on the IEEE 33-bus system with high PV penetration. Simulation results demonstrate that the proposed approach achieves a 22.2% reduction in operational cost, significantly improves voltage profiles, and reduces performance variability under uncertainty compared to rule-based control and optimal power flow methods. The integration of DRL and Digital Twin enables adaptive and robust decision-making, making the framework suitable for modern distribution networks with high renewable penetration.

Performance Evaluation and Energy-Saving Prioritization of Thermal Oil Heater in Plastic Manufacturing Plant

Arya Bima Putra, Faiz Husnayain — Tue, 30 Jun 2026 00:00:00 +0700

Heating systems are one of the main contributors to energy consumption in industry. This study was conducted on the Thermal Oil Heater (TOH) system at PT XYZ, a plastic manufacturing company in West Java, to evaluate TOH performance using the direct method, linear regression modeling, and the Energy Performance Indicator, as well as to identify and prioritize energy-saving opportunities in the TOH system using the Analytic Hierarchy Process. The results of the direct method show that the average TOH efficiency was 65.78%. The linear regression equation was η = 1246.889 - 0.2428X₁ - 4.2929X₂, where gas flow rate and thermal oil inlet temperature significantly affected TOH efficiency. The EnPI method produced a baseline value of 1.58. The identified energy-saving potential was 26%. Based on the AHP analysis, the priority order of the criteria is as follows: energy-saving potential, ease of implementation, impact on operations and production, and investment cost. The priority order of the alternatives is as follows: optimization of gas fuel consumption rate, improvement of monitoring and control of operating parameters, improvement of the TOH system maintenance program, more efficient operating temperature control, increasing the temperature difference, burner combustion optimization, periodic cleaning of heat transfer pathways, and improvement of pipe and TOH body insulation.

Particle Swarm Optimization for Risk-Aware Distributed Generation Planning in Distribution Networks

ton trieu — Wed, 01 Jul 2026 07:08:59 +0700

High penetration of distributed generation (DG) improves distribution network performance but may increase voltage rise and reverse power flow, threatening secure operation. This paper proposes a Particle Swarm Optimization (PSO)-based DG planning framework that simultaneously minimizes active power loss, mitigates voltage risk, suppresses reverse power flow, and maximizes DG penetration. A normalized multi-objective function integrates these criteria into a unified optimization model, while penalty functions enforce network operating constraints. The proposed method is validated on the 33-bus and 69-bus radial distribution systems and compared with GSA, MVMO-SH, and conventional PSO. For the 33-bus system, the proposed approach reduces active power loss by 36.47%, increases the minimum bus voltage from 0.913 p.u. to 0.964 p.u., and limits reverse power flow to 13.2 kW. For the 69-bus system, it achieves a 61.92% reduction in power loss, improves the minimum voltage from 0.909 p.u. to 0.972 p.u., and decreases reverse power flow to 18.9 kW while providing the highest DG penetration among the compared methods. The results demonstrate that explicitly considering voltage risk and reverse power flow enables more effective and reliable DG planning for active distribution networks.

Prototype Design of a Prepaid KWh Meter Keypad Pusher with the Internet of Things to Refill Tokens

Muhamad Nur Mahmudi, Anhar, Mukmin Maulana Latin — Wed, 01 Jul 2026 00:00:00 +0700

Prepaid kWh meters require manual token entry via a physical keypad, which limits operational flexibility, particularly for remote or unattended installations. Delayed token input may result in unintended power interruptions. This paper presents the design and implementation of an Internet of Things (IoT)-based keypad actuation system that enables remote token entry without modifying or replacing the existing prepaid meter. The proposed system employs servo motor actuators driven by an ESP8266 microcontroller and utilizes a Firebase cloud server for data exchange between the embedded device and an Android-based user interface developed using Kodular. An ESP32-CAM module is integrated for real-time visual monitoring of the meter display. Experimental evaluation shows that the prototype achieves a keypad actuation accuracy of 98.33% and a 100% success rate in token charging. Voltage measurements indicate system deviations below 5%, confirming stable electrical operation. Measured communication latency includes an average delay of 1706.63 ms from the graphical user interface to the cloud server and 483.6 ms from the server to the microcontroller. The results demonstrate that the proposed system provides a reliable and cost-effective solution for remote prepaid electricity token management.

Behavior Retrieval plus Response Generation for Interpretable Conversational Personalized Recommendation

Qi Xin — Fri, 03 Jul 2026 12:28:07 +0700

This paper presents an empirical study of interpretable conversational personalized recommendation built from two tightly coupled layers: a behavior retrieval layer for candidate ranking and a response generation layer for grounded recommendation wording. The study is motivated by recent large language model surveys in recommendations, but the implementation remains deterministic so that candidate selection, evidence extraction, and wording can be inspected separately. We used the 2010-2011 partition of Online Retail II for transaction-grounded next-basket recommendation and a retail/service subset of the Bitext customer-support 27K corpus for conversational response generation. The behavior retrieval layer combined popularity, repeat purchase memory, recency-weighted memory, item-to-item collaborative filtering, and user-neighborhood scoring. The response layer compared intent-majority, TF-IDF retrieval, intent-conditioned retrieval, and a constrained fusion verbalizer. On 1,999 test users, the hybrid behavior retriever achieved Hit@10 = 0.7282, Recall@10 = 0.1982, NDCG@10 = 0.2855, and MRR@10 = 0.4467, outperforming the strongest baseline RecencyRepeat by 2.03% on Hit@10 and 6.34% on NDCG@10. On 1,989 Bitext test utterances, intent-conditioned retrieval produced the best standalone verbalization quality with BLEU-4 = 0.1699, ROUGE-L = 0.3431, intent accuracy = 0.9915, and slot F1 = 0.8239. Grounded recommendation responses reached Top1Hit = 0.3252, HistoryGroundRate = 1.0000, and BundleValidity = 0.9980. The results show that retrieval-first conversational recommendation can combine ranking accuracy, grounding faithfulness, and low-overhead response realization without relying on unconstrained generation.