Zhengyang Ling
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Zhengyang Ling has a total of 33 co-authors. Here are the top co-authors by number of shared publications:
Alan Thorne
4 shared publicationsDuncan McFarlane
3 shared publications- Duncan McFarlane
University of Cambridge
2 shared publications Gregory Hawkridge
2 shared publicationsSam Brooks
2 shared publications- Svetan Ratchev
Institute for Advanced Manufacturing, University of Nottingham, Nottingham, UK
2 shared publications
Latest Publications
- Open Access
- 6 November 2026
- Published by Springer Nature in The International Journal of Advanced Manufacturing Technology
- p. 1-23
Evaluating the use of LLM-based multi-agent systems for internal information sharing and access within manufacturing
Internal information sharing in manufacturing remains constrained by data silos, heterogeneous systems, poor data quality, and limited interoperability across ISA-95 levels. While Large Language Models (LLMs) are increasingly proposed for manufacturing information access and for integrating information with operational technology, there is limited empirical evidence comparing their performance with unassisted human users on representative manufacturing tasks. This paper evaluates a modular LLM-based multi-agent system (MAS) for retrieving and synthesising information across heterogeneous data sources from a 3D-printing factory demonstrator. The system was benchmarked against human participants and other LLM systems (Retrieval-Augmented Generation, Single Agent, Zero-Shot, and rule-based deterministic system) on traceability and certification tasks involving packing lists, relationship-tracking records, machine logs, location-tracking data, and Authorised Release Certificates (ARCs; FAA Form 8130-3). Three task types under four data-quality conditions were tested, including: unaltered baseline data, whitespace injection, truncated identifiers, and severed relational links. Under the bounded experimental conditions, the highest-performing LLMs achieved higher mean accuracy and shorter completion times than the average unassisted participant, with Claude Sonnet-4 achieving 94.1% mean accuracy compared with 62.2% for humans. However, this advantage was conditional: some smaller models fell below the human average, and the best human participants outperformed all models. The MAS system was robust to whitespace errors and severed relational links, but vulnerable to truncated identifiers and verbose prompt scaffolding. These findings offer a preliminary indication that multi-agent LLM systems can support bounded manufacturing information-access tasks as a decision-support layer, but require careful model selection, tool design, data validation, prompt design, and human oversight.
- 6 November 2026
- Published by Springer Nature in The International Journal of Advanced Manufacturing Technology
- p. 1-23
Open AccessEvaluating the use of LLM-based multi-agent systems for internal information sharing and access within manufacturing
Internal information sharing in manufacturing remains constrained by data silos, heterogeneous systems, poor data quality, and limited interoperability across ISA-95 levels. While Large Language Models (LLMs) are increasingly proposed for manufacturing information access and for integrating information with operational technology, there is limited empirical evidence comparing their performance with unassisted human users on representative manufacturing tasks. This paper evaluates a modular LLM-based multi-agent system (MAS) for retrieving and synthesising information across heterogeneous data sources from a 3D-printing factory demonstrator. The system was benchmarked against human participants and other LLM systems (Retrieval-Augmented Generation, Single Agent, Zero-Shot, and rule-based deterministic system) on traceability and certification tasks involving packing lists, relationship-tracking records, machine logs, location-tracking data, and Authorised Release Certificates (ARCs; FAA Form 8130-3). Three task types under four data-quality conditions were tested, including: unaltered baseline data, whitespace injection, truncated identifiers, and severed relational links. Under the bounded experimental conditions, the highest-performing LLMs achieved higher mean accuracy and shorter completion times than the average unassisted participant, with Claude Sonnet-4 achieving 94.1% mean accuracy compared with 62.2% for humans. However, this advantage was conditional: some smaller models fell below the human average, and the best human participants outperformed all models. The MAS system was robust to whitespace errors and severed relational links, but vulnerable to truncated identifiers and verbose prompt scaffolding. These findings offer a preliminary indication that multi-agent LLM systems can support bounded manufacturing information-access tasks as a decision-support layer, but require careful model selection, tool design, data validation, prompt design, and human oversight.
- Open Access
- 25 June 2026
- Published by Elsevier in SSRN Electronic Journal
Decomposing Manufacturing Tracking
Tracking systems support manufacturers with order management, product tracking, asset control, and other related functions, as well as providing detailed traceability data. However, developing and deploying effective tracking systems is often costly due to the need for customisation and the complexity of integrating diverse technologies and data sources, frequently resulting in bespoke solutions or the deployment of multiple tracking systems with overlapping functions. An analysis of existing research identified 10 core tracking functions currently used in manufacturing, including order, product, part, job, operation, resource, material, worker, asset, and logistics tracking. By analysing data from the literature and user requirements for these core functions, common data types were identified. A clustering analysis of these data types showed that they can be grouped into three elementary tracking functions: location, activity, and relationship tracking. This enables modular design, reduces system complexity, and minimises overlap between software functions. The effectiveness of the proposed decomposition was assessed using a laboratory system which emulates the operations of a 3D printing factory. All tracking and traceability requirements were met by the system, which consisted of elementary tracking solutions, although 7 of the 23 requirements fell below the 0.95 accuracy threshold. The results support the feasibility of the decomposition and highlight its potential to simplify the development of manufacturing tracking systems.
- 25 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessDecomposing Manufacturing Tracking
Tracking systems support manufacturers with order management, product tracking, asset control, and other related functions, as well as providing detailed traceability data. However, developing and deploying effective tracking systems is often costly due to the need for customisation and the complexity of integrating diverse technologies and data sources, frequently resulting in bespoke solutions or the deployment of multiple tracking systems with overlapping functions. An analysis of existing research identified 10 core tracking functions currently used in manufacturing, including order, product, part, job, operation, resource, material, worker, asset, and logistics tracking. By analysing data from the literature and user requirements for these core functions, common data types were identified. A clustering analysis of these data types showed that they can be grouped into three elementary tracking functions: location, activity, and relationship tracking. This enables modular design, reduces system complexity, and minimises overlap between software functions. The effectiveness of the proposed decomposition was assessed using a laboratory system which emulates the operations of a 3D printing factory. All tracking and traceability requirements were met by the system, which consisted of elementary tracking solutions, although 7 of the 23 requirements fell below the 0.95 accuracy threshold. The results support the feasibility of the decomposition and highlight its potential to simplify the development of manufacturing tracking systems.
- Open Access
- 22 June 2026
- Published by IGLC
- p. 1499-1510
A minimal-data toolkit for diagnosing loss of flow resilience in interior fit-out work
This paper proposes and empirically tests a minimal flow performance measurement toolkit for construction fit-out works, designed to characterise flow stability, continuity, and vulnerability to schedule disruption using only routinely collected progress observations. The toolkit operates across activity, trade, and …
- 22 June 2026
- Published by IGLC
- p. 1499-1510
Open AccessA minimal-data toolkit for diagnosing loss of flow resilience in interior fit-out work
This paper proposes and empirically tests a minimal flow performance measurement toolkit for construction fit-out works, designed to characterise flow stability, continuity, and vulnerability to schedule disruption using only routinely collected progress observations. The toolkit operates across activity, trade, and …
- Open Access
- 15 July 2025
- Published by Elsevier in Robotics and Computer-Integrated Manufacturing
- Vol. 97
Vision-based extraction of industrial information from legacy Programmable Logic Controllers
- 15 July 2025
- Published by Elsevier in Robotics and Computer-Integrated Manufacturing
- Vol. 97
Open AccessVision-based extraction of industrial information from legacy Programmable Logic Controllers
- 10 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Services Computing
- Vol. 17 (5) , 2139-2150
Cross-Edge Orchestration of Serverless Functions With Probabilistic Caching
Serverless edge computing adopts an event-based paradigm that provides back-end services and dynamically provisions resources as needed, resulting in efficient resource utilization. To improve the end-to-end latency and revenue, service providers need to optimize the number and placement of serverless containers while considering the system cost (i.e., latency cost and container running cost) incurred by the provisioning. The particular reason for this circumstance is that frequently creating and destroying containers not only increases the system cost but also degrades the time responsiveness due to the cold-start process. Function caching is a common approach to mitigate the coldstart issue. However, function caching requires extra hardware resources and hence incurs extra system costs. Furthermore, the dynamic and bursty nature of serverless invocations remains an under-explored area. Hence, it is vitally important for service providers to conduct a context-aware request distribution and container caching policy for serverless edge computing. In this paper, we study the request distribution and container caching problem in serverless edge computing. We prove the proposed problem is NP-hard and hence difficult to find a global optimal solution. We jointly consider the distributed and resourceconstrained nature of edge computing and propose an optimized request distribution algorithm that adapts to the dynamics of serverless invocations with a theoretical performance guarantee. Also, we propose a context-aware probabilistic caching policy that incorporates a number of characteristics of serverless invocations. Via simulation and implementation results, we demonstrate the superiority of the proposed algorithm by outperforming existing caching policies in terms of the overall system cost and cold-start frequency by up to 62.1% and 69.1%, respectively.
- 10 May 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Services Computing
- Vol. 17 (5) , 2139-2150
Cross-Edge Orchestration of Serverless Functions With Probabilistic Caching
Serverless edge computing adopts an event-based paradigm that provides back-end services and dynamically provisions resources as needed, resulting in efficient resource utilization. To improve the end-to-end latency and revenue, service providers need to optimize the number and placement of serverless containers while considering the system cost (i.e., latency cost and container running cost) incurred by the provisioning. The particular reason for this circumstance is that frequently creating and destroying containers not only increases the system cost but also degrades the time responsiveness due to the cold-start process. Function caching is a common approach to mitigate the coldstart issue. However, function caching requires extra hardware resources and hence incurs extra system costs. Furthermore, the dynamic and bursty nature of serverless invocations remains an under-explored area. Hence, it is vitally important for service providers to conduct a context-aware request distribution and container caching policy for serverless edge computing. In this paper, we study the request distribution and container caching problem in serverless edge computing. We prove the proposed problem is NP-hard and hence difficult to find a global optimal solution. We jointly consider the distributed and resourceconstrained nature of edge computing and propose an optimized request distribution algorithm that adapts to the dynamics of serverless invocations with a theoretical performance guarantee. Also, we propose a context-aware probabilistic caching policy that incorporates a number of characteristics of serverless invocations. Via simulation and implementation results, we demonstrate the superiority of the proposed algorithm by outperforming existing caching policies in terms of the overall system cost and cold-start frequency by up to 62.1% and 69.1%, respectively.
- Open Access
- 24 November 2023
- Published by IOS Press
- p. 99-109
Design of a Demonstrator Environment for Investigating Multi-Factory Production and Operation Challenges
Demonstrators, testbeds and learning factories enable researchers to investigate important manufacturing challenges and to trial solutions without disrupting industrial production facilities. In this way, solutions and systems can be developed close to a ‘production-ready’ state prior to industrial deployment. This paper reviews demonstrators and testbeds developed for smart manufacturing in the last 20 years. A key observation is that such demonstrators have predominantly focused on emulating single or multiple closely connected operations. Such developments reflect the activities of a single production facility and/or organisation. In contrast, there are few reports on demonstrators which seek to replicate the behaviour and challenges associated with multi-site factories or integration with existing legacy factory systems. To address this gap, a multi-operation demonstrator has been created. The demonstrator aims to replicate coordinated production between multiple small manufacturing sites and provides a testbed to investigate operational challenges. The current demonstrator, the research investigated, and the direction of future research proposed are outlined.
- 24 November 2023
- Published by IOS Press
- p. 99-109
Open AccessDesign of a Demonstrator Environment for Investigating Multi-Factory Production and Operation Challenges
Demonstrators, testbeds and learning factories enable researchers to investigate important manufacturing challenges and to trial solutions without disrupting industrial production facilities. In this way, solutions and systems can be developed close to a ‘production-ready’ state prior to industrial deployment. This paper reviews demonstrators and testbeds developed for smart manufacturing in the last 20 years. A key observation is that such demonstrators have predominantly focused on emulating single or multiple closely connected operations. Such developments reflect the activities of a single production facility and/or organisation. In contrast, there are few reports on demonstrators which seek to replicate the behaviour and challenges associated with multi-site factories or integration with existing legacy factory systems. To address this gap, a multi-operation demonstrator has been created. The demonstrator aims to replicate coordinated production between multiple small manufacturing sites and provides a testbed to investigate operational challenges. The current demonstrator, the research investigated, and the direction of future research proposed are outlined.
- 3 June 2022
- Published by Springer Nature
- p. 277-289
Visualisation on a Shoestring: A Low-Cost Approach for Building Visualisation Components of Industrial Digital Solutions
- 3 June 2022
- Published by Springer Nature
- p. 277-289
Visualisation on a Shoestring: A Low-Cost Approach for Building Visualisation Components of Industrial Digital Solutions
- 3 June 2022
- Published by Springer Nature
- p. 263-275
A Graphical Environment to Support the Development of Affordable Digital Manufacturing Solutions
- 3 June 2022
- Published by Springer Nature
- p. 263-275
A Graphical Environment to Support the Development of Affordable Digital Manufacturing Solutions