Sustainable and Intelligent Public Facility Failure Management System Based on Large Language Models

TL;DR

The paper addresses the challenge of managing intelligent devices in public facilities, focusing on libraries where data are heterogeneous and real-time maintenance is critical. It introduces an LLM-based framework that semantically fuses multi-source data into a local knowledge base to identify and prevent device failures, while providing personalized maintenance guidance and procurement insights. A library-based prototype using a multimodal dataset and the Llama 3.2-Vision foundation demonstrates proactive failure prevention and potential cost savings. The work envisions expanding to additional public facilities and integrating IoT cybersecurity for threat detection and response, further strengthening resilience and service quality.

Abstract

This paper presents a new Large Language Model (LLM)-based Smart Device Management framework, a pioneering approach designed to address the intricate challenges of managing intelligent devices within public facilities, with a particular emphasis on applications to libraries. Our framework leverages state-of-the-art LLMs to analyze and predict device failures, thereby enhancing operational efficiency and reliability. Through prototype validation in real-world library settings, we demonstrate the framework's practical applicability and its capacity to significantly reduce budgetary constraints on public facilities. The advanced and innovative nature of our model is evident from its successful implementation in prototype testing. We plan to extend the framework's scope to include a wider array of public facilities and to integrate it with cutting-edge cybersecurity technologies, such as Internet of Things (IoT) security and machine learning algorithms for threat detection and response. This will result in a comprehensive and proactive maintenance system that not only bolsters the security of intelligent devices but also utilizes machine learning for automated analysis and real-time threat mitigation. By incorporating these advanced cybersecurity elements, our framework will be well-positioned to tackle the dynamic challenges of modern public infrastructure, ensuring robust protection against potential threats and enabling facilities to anticipate and prevent failures, leading to substantial cost savings and enhanced service quality.

Figures (17)

• Figure 1: The typical public facilities where smart devices are widely used to provide various basic services.
• Figure 2: The structure of LLM-based Smart Device Failure Management, where the semantic integration of multi-source heterogeneous data is achieved, and it is combined with the efficient understanding and processing capabilities of LLMs.
• Figure 3: The schema of the considered smart devices deployed in the library for simulation.
• Figure 4: The schematic layout of smart devices in the corridor.
• Figure 5: The image of Self-service Borrowing and Returning Machine #1 taken from the front by Surveillance Camera #1 .