Scalable Machine Learning Architectures for IPA-Driven Maintenance Task Allocation in Large-Scale Building Portfolios

Huang, Zhimeng; Liu, Xiaodong; Romdhani, Imed; Shih, Chi-Sheng

Abstract

This research presents a groundbreaking approach to Building Maintenance Management (BMM) by introducing an Intelligent Process Automation (IPA)-Driven Building Maintenance Management (IBMM) model. This innovative model harnesses the synergies between Artificial Intelligence (AI), Machine Learning (ML), and Internet of Things (IoT) technologies to transition from reactive to proactive and predictive building maintenance strategies. The study highlights the critical gap in current BMM practices—the absence of intelligent systems for anticipating and addressing maintenance issues before they escalate. Through an extensive literature review, the transformative potential of AI and IoT for enhancing building maintenance management within smart cities is explored, establishing a foundation for the IBMM model's application.

The core of this research lies in its novel application of scalable machine learning architectures to automate and optimize maintenance task allocation in large-scale building portfolios. The practicality of the IBMM model is demonstrated via a proof of concept (POC) in an industrial setting, evidencing its capacity to improve efficiency, reduce costs, and bolster sustainability in building maintenance operations. The model epitomizes a paradigm shift in BMM by integrating IPA, which combines AI and ML, facilitating automated, intelligent decision-making and task allocation. Among its advancements, the IBMM model introduces enhanced predictive maintenance through real-time data analysis, adaptive learning and optimization, automated decision-making, and human-machine collaboration, contributing to energy efficiency and alignment with smart city objectives.

The paper delineates the methodology, design, and implementation of a machine learning model for engineer task assignments, culminating in a case study that validates the model's efficacy. This research not only signifies a significant advancement in BMM by leveraging IPA technologies for autonomous process refinement but also proposes a unique IPA-driven procedure that incorporates IoT technology and a novel smart device fixer to guide BMM processes.

Anticipated outcomes include more accurate maintenance scheduling, cost efficiency, enhanced performance, and the fostering of a collaborative community through an open online documentation platform for BMM. Looking forward, the research aims to refine the IBMM model further by exploring advanced AI algorithms for more precise predictive maintenance and integrating real-time data analytics and IoT networks for improved maintenance strategy responsiveness. This work pioneers a smarter, more efficient, and sustainable approach to building maintenance, marking a new era in the management of urban infrastructure.

Citation

Huang, Z., Liu, X., Romdhani, I., & Shih, C.-S. (2024, August). Scalable Machine Learning Architectures for IPA-Driven Maintenance Task Allocation in Large-Scale Building Portfolios. Presented at The 7th International Conference on Information Science and Systems (ICISS 2024), Edinburgh, UK

• Presentation Conference Type: Conference Paper (published)
• Conference Name: The 7th International Conference on Information Science and Systems (ICISS 2024)
• Start Date: Aug 14, 2024
• End Date: Aug 16, 2024
• Acceptance Date: Apr 17, 2024
• Online Publication Date: Jan 31, 2025
• Publication Date: 2025
• Deposit Date: May 28, 2024
• Publicly Available Date: Jan 31, 2025
• Publisher: Association for Computing Machinery (ACM)
• Peer Reviewed: Peer Reviewed
• Pages: 170-178
• Book Title: ICISS '24: Proceedings of the 2024 7th International Conference on Information Science and Systems
• ISBN: 9798400717567
• DOI: https://doi.org/10.1145/3700706.3700734
• Keywords: Building Maintenance Management (BMM); Machine Learning (ML); Intelligent Process Automation (IPA); Decision Tree
• Public URL: http://researchrepository.napier.ac.uk/Output/3651290
• Publisher URL: https://dl.acm.org/conference/iciss/proceedings
• External URL: https://iciss.org/