一句话：提出集成 LLM 的生成式 AI 框架，实现建筑运维自动化与智能虚拟助手交互。
传统智能家居助手主要作为语音激活的控制系统，适应性与情境理解能力有限；传统建筑 AI 研究则多依赖预定义规则，难以在动态建筑环境中实时决策。本文提出一种新的生成式 AI 驱动框架，集成大语言模型（LLM），构建面向建筑基础设施的智能虚拟助手与运维自动化系统。该系统通过分析实时人员占用模式，并依据预设的舒适度阈值自动调节环境条件，同时借助 LLM 驱动的虚拟助手实现顺畅的人—建筑交互。作者在配备智能电器的真实建筑中实现原型系统进行验证，评估聚焦于 AI 系统的准确性、可靠性与可扩展性。结果表明，该原型系统能够自主调节建筑环境、优化能耗，并为建筑运维任务提供智能辅助。
Conventional AI-powered smart home assistants primarily function as voice-activated control systems with limited adaptability and contextual understanding. Similarly, while traditional artificial intelligence has advanced autonomous building research, it often relies on predefined rules and struggles with real-time decisionmaking in dynamic building environments. This paper introduces a novel Generative AI-driven framework that integrates Large Language Models (LLMs) to create a smart generative AI-based virtual assistant and an operation automation system for building infrastructure. The AI systems autonomously manage building operations by analyzing real-time occupancy patterns and adjusting environmental conditions based on predefined comfort thresholds. The proposed system also facilitates seamless human-building interaction through an LLM-powered virtual assistant. The framework is validated through a prototype implementation in a real-world building equipped with smart appliances, with evaluations focusing on the AI systems' accuracy, reliability, and scalability. The findings demonstrate that the prototype system can autonomously adjust building conditions, optimize energy usage, and provide intelligent assistance for building operation tasks.