Description
Efficient knowledge management is increasingly critical in the energy sector, where timely, accurate information can drive sustainable practices and operational efficiency [1]. However, traditional knowledge management systems often fall short in meeting the complex and evolving needs of the energy domain, particularly in integrating renewable energy sources [2]. Current systems lack flexibility and struggle to support the dynamic, domain-specific queries essential for decision-making in renewable energy management [3].To address this gap, our research introduces the Advanced Agent Architecture (AAA), an innovative framework that enhances information retrieval by combining RetrievalAugmented Generation (RAG) techniques with a localised knowledge base (LKB) and real-time web search capabilities. This design enables AAA to access both domainspecific data and external sources, offering robust support for complex queries within the energy sector. AAA’s architecture integrates advanced methods, including query routing [4], fallback mechanisms [5], and self-correction [6], to ensure accurate and contextually relevant responses [7].
Through a case study focused on a solar power system, we evaluated AAA’s performance on tasks like technical troubleshooting, interpreting error codes, and maintenance planning informed by live weather data. Findings demonstrate that AAA delivers significantly higher accuracy than traditional Large Language Models (LLMs) like Llama 3, particularly in scenarios requiring real-time data access. Despite these advantages, AAA’s increased energy consumption suggests areas for further optimization.
This research underscores AAA’s potential to advance sustainable knowledge management practices in the energy sector, supporting informed decision-making. Future work will target enhancing energy efficiency and processing speed to make AAA a practical tool for broad adoption, aligning with industry goals for sustainability and operational efficacy.
| Period | 29 Nov 2024 |
|---|---|
| Event title | Doctoral School Multidisciplinary Symposium (DSMS) 2024 |
| Event type | Conference |
| Location | Glasgow, United KingdomShow on map |
| Degree of Recognition | Regional |