Crossover-BPSO Driven Multi-Agent Technology for Managing Local Energy Systems
Hafiz Majid Hussain, Ashfaq Ahmad, Pedro H. J. Nardelli · 2025
Read Crossover-BPSO Driven Multi-Agent Technology for Managing Local Energy Systems by Hafiz Majid Hussain, Ashfaq Ahmad, Pedro H. J. Nardelli (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-09832.txt.
Introduction
Hafiz Majid Hussain, Ashfaq Ahmad, and Pedro Nardelli present a 2025 study on managing local energy systems with software agents. Their work combines a crossover form of binary particle swarm optimization with a layered agent design for a setup connected to the main grid. One lead agent runs the planning while smaller agents report local conditions.
Inside, the lead agent works to cut energy buying costs, limit battery wear, and reduce delays in scheduling loads. The smaller agents send status updates and receive action plans in return. In simulations the authors report lower average system costs than the methods they compared against, while still meeting user demand and service quality.
Catalog record
| CREATOR | Hafiz Majid Hussain, Ashfaq Ahmad, Pedro H. J. Nardelli |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 24 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-09832
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Crossover-BPSO Driven Multi-Agent Technology for Managing Local Energy Systems Hafiz Majid Hussain, Ashfaq Ahmad, Pedro H. J. Nardelli Abstract This article presents a new hybrid algorithm, crossover binary particle swarm optimization (crBPSO), for allocating resources in local energy systems via multi-agent (MA) technology. Initially, a hierarchical MA-based architecture in a grid-connected local energy setup is presented. In this architecture, task specific agents operate in a master-slave manner. Where, the master runs a well-formulated optimization routine aiming at minimizing costs of energy procurement, battery degradation, and load scheduling delay. The slaves update the master on their current status and receive optimal action plans accordingly. Simulation results demonstrate that the proposed algorithm outperforms selected existing ones by 21% in terms average energy system costs while satisfying customers’ energy demand and maintaining the required quality of service. PubID: pubid: 979-8-3503-9042-1/24/$31.00 ©2024 IEEE I Introduction Local energy systems (LESs) are gradually transitioning from centralized and high-emission systems to hybrid and eco-friendly ones, mainly via DER (distributed energy resources) integration such as renewable energy sources (RESs), energy storage units, and demand response (DR) technologies [ 1 ] , [ 2 ] . However, these integrations may face several challenges like the intermittency in RESs output power generation, supply uncertainty, and dynamic power system states posing stability implications [ 3 ] . In order to address these challenges, a multi-agent system (MAS) technology can be adopted for managing complex DER interactions within LES through a hierarchy of smart agents operating either independently or in collaboration to achieve either a common objective ( e.g. , resource optimization) or a set of objectives (like energy efficiency, grid stability, customer satisfaction, grid resilience). Hence, an advanced MAS technology driven by intelligent management and control is essential for ensuring green and cost-effective LES solutions [ 4 ] . Previously, various energy management approaches have been explored to solve energy management problems in MAS to achieve economical solutions for energy users in LESs. For instance, authors in [ 5 ] employed MAS technology to reduce LES’s operational expenses using a rule-based real-time corrective algorithm and DR strategy. The articles [ 6 ] and [ 7 ] presented MAS-based residential EMS (energy management system) to achieve multiple objectives including energy optimization, user satisfaction cost, and carbon emission reduction using distributed optimization and a greedy randomized adaptive search procedure. The authors in [ 8 ] proposed a novel MAS-based EMS for energy internet-of-things (EIoT) devices to accomplish minimal energy loss and cost reduction by employing Lagrangian multiplier method. Similarly, the authors in [ 9 ] proposed an intelligent MAS-based EMS architecture for smart buildings to achieve energy optimization and cost minimization via machine learning approaches (k-nearest neighbors algorithm and support vector machines). Similarly, refs. [ 3 ] and [ 10 ] presented a multi-layer MAS-based architecture to enhance reliable coordination via communication among various MAs such as DERs, efficient EMS, and cost minimization. In other similar works, EMS architectures were proposed to achieve specific objectives like optimization of energy usage via peak-to-average (PAR) reduction through heuristic algorithms in [ 11 ] and [ 12 ] , and reduction of average system cost in [ 13 ] via Lyapunov optimization. From an optimization standpoint, the approaches in in [ 5 , 6 , 7 , 8 ] have shown good performance in particular problem contexts, including situations with predefined rules, large-scale scenarios, and constraint-based issues. Nevertheless, the lack of flexibility and adaptability poses significant challenges when employing these approaches. To address these, heuristic algorithms (like genetic algorithm (GA), binary particle swarm optimization (BPSO), differential evolution (DE) and harmony search (HS)) are more effective in rapidly exploring complex search spaces to find near-optimal solutions, and flexibly adapting to diverse problem instances [ 12 , 13 , 14 ] . Fig. 1 : MA-based LES architecture involving SRAs (loads and PV source), MRAs (battery storage and ESP), UA (energy router) and the external power grid. Based on our comprehensive review of existing literature [ 1 , 3 , 4 , 5 , 6 , 7 , 8 , 9 , 10 , 11 , 12 , 13 ] above, it is evident that most prior studies have concentrated on exploring EMS solutions, either incorporating MAS technology or not. Despite their benefits, the following research gaps have been identified. First : despite the diversity in EMS components and objectives, the agents designed and employed were homogeneous. For example, using either a group of simple reflex-based agents (SRAs) or model-based agents (MRAs) for maintaining a balanced energy load across a diverse set of devices and systems within a LES by dynamically adjusting electricity consumption in response to energy prices. This homogeneity limits the full potential of MAS technology, thereby reducing its effectiveness for EMS applications in LES. Second : within the two main families of heuristic optimization methods, genetic evolution-based approaches lack elements of social learning, while swarm movement-based approaches lack genetic diversity. For example, the exploration phases of GA and DE involve diversity through a non-directed genetic recombination process, and that of PSO involve directed velocity-based position updates without taking into account recombination-based diversity [ 14 ] . As a result, the exploratory phase in these algorithms is constrained, leading to convergence at local optima. To address the aforementioned research gaps, this work introduces a hierarchical MA technology-based architecture powered by our newly proposed crBPSO algorithm for EMS applications in LES. Then, simulations are carried to validate the relative effectiveness of the proposed work in terms of selected performance metrics. The main contributions and benefits of this research can be summarized as follows. • A novel hierarchical MA technology-based architecture is introduced, featuring heterogeneous agents (SRAs, MRAs and utility agent (UA)) designed and employed to account for diverse EMS components and objectives. • An energy transaction model is developed for energy trade between energy consumers and energy service provider (ESP). This model enables cost-effective solutions for energy consumers while maintaining demand-supply balance. • A new hybrid heuristic algorithm, called crossover-BPSO ( crBPSO ), is proposed. It integrates crossover from genetic algorithms into the velocity update equation of BPSO, enhancing exploration abilities and enabling deeper search space exploration. The rest of the paper is organized as follows. Section II presents the multi-gent system model, formulates the problem, and presents the crBPSO algorithm. Section III discusses the simulation results, and section IV concludes the paper. References are provides at the end. II Multi-Agent System Model Consider a MA-based architecture shown in figure 1 that encompasses three types of agents: simple reflex agents (SRAs), model-based-reflex agents (MRAs), and a utility-based agent (UA). The upcoming sections provide details on the explicit models of MAs along with their hierarchical functioning to devise optimal energy packet transaction schedules between ESP and prosumers. II-A Simple-Reflex Agents In this work, smart home loads and PV source are modelled as SRAs, as both undertake rule-based decisions depending only on the current percept. The smart energy community ( i.e. , LES) shown in figure 1 consists of an array of smart homes b\in\{1,2,3,\;...,\;B\} such that each smart home has multiple IoT-enabled loads l\in\{1,2,3,..,L\} modelled as SRAs. The energy consumed by these agents is quantized in the form of energy packets at discrete time slots t_{s}\in\{0,1,2,3,...T_{s}-1\} , such