Helios: A Foundational Language Model for Smart Energy Knowledge Reasoning and Application

2025-12-22

AI Summary

Achieving carbon neutrality demands intelligent energy systems powered by large language models (LLMs) that possess domain-specific knowledge and awareness of physical constraints—capabilities absent in general-purpose LLMs due to their lack of energy-domain expertise and engineering alignment. To address this gap, we propose Helios, the first domain-specialized LLM for intelligent energy systems. We introduce Enersys, a novel multi-agent collaborative data engineering framework that systematically generates three high-quality resources: EnerBase (a domain-knowledge-enhanced knowledge base), EnerInstruct (an instruction-tuning dataset), and EnerReinforce (an RLHF alignment dataset grounded in energy physics and standards). We further release EnerBench, the first comprehensive evaluation benchmark for intelligent energy LLMs. Through domain-knowledge-augmented pretraining, supervised fine-tuning, and physics-informed RLHF, Helios achieves substantial improvements over general-purpose baselines in energy knowledge comprehension, engineering task accuracy, and compliance with industry standards.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Natural Language Processing: (Large) Language Models

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information

Abstract

In the global drive toward carbon neutrality, deeply coordinated smart energy systems underpin industrial transformation. However, the interdisciplinary, fragmented, and fast-evolving expertise in this domain prevents general-purpose LLMs, which lack domain knowledge and physical-constraint awareness, from delivering precise engineering-aligned inference and generation. To address these challenges, we introduce Helios, a large language model tailored to the smart energy domain, together with a comprehensive suite of resources to advance LLM research in this field. Specifically, we develop Enersys, a multi-agent collaborative framework for end-to-end dataset construction, through which we produce: (1) a smart energy knowledge base, EnerBase, to enrich the model's foundational expertise; (2) an instruction fine-tuning dataset, EnerInstruct, to strengthen performance on domain-specific downstream tasks; and (3) an RLHF dataset, EnerReinforce, to align the model with human preferences and industry standards. Leveraging these resources, Helios undergoes large-scale pretraining, SFT, and RLHF. We also release EnerBench, a benchmark for evaluating LLMs in smart energy scenarios, and demonstrate that our approach significantly enhances domain knowledge mastery, task execution accuracy, and alignment with human preferences.

Problem

• Addresses smart energy domain knowledge gaps in LLMs
• Enhances model alignment with engineering constraints and standards
• Improves accuracy in domain-specific reasoning and tasks

Innovation

• Multi-agent collaborative framework for dataset construction
• Domain-specific knowledge base and instruction fine-tuning datasets
• Large-scale pretraining with SFT and RLHF alignment