Adaptive network analytics for managing complex shop-floor logistics systems
Introduction
Like bloodstream for living beings, internal logistics for production systems provides the essential circulation of material. Even though this service adds no direct value, it has a crucial impact on production, almost according to every key performance indicator (KPI). Ideally, internal logistics sees if all parts, components, materials, fixtures, tools and auxiliary equipment are available just in the right time and quantity for production. It also instantly sets free sites of production from their results, by-products, scrap, and unneeded equipment. In reality, however, internal logistics is often accountable for blocking production by shortage or accumulation of material, causing loss of time and resources.
Modern automated guided vehicles (AGVs) are driverless, free-ranging transport devices with localization and autonomous control faculties [1]. AGVs operate typically in a fleet, carrying loads of multiple types and cardinality. Recently, their industrial application has proliferated [2,3], and an even more significant expansion of their use is to be expected, as line production is giving headway to reconfigurable or changeable manufacturing [4].
This work was motivated to provide AGV service for complex, large-scale production where even though processing times and quality are fraught by uncertainties, interrupts and re-entrant work are frequent, machine utilization is of primary importance. Adapting to changing workload, responding to changes on the shop-floor quickly, and scaling up with the extension of the production facilities were also key requirements. All these conditions preclude the advance joint scheduling of production and logistics tasks [5]. Instead, as in most of the cases, production should generate tasks for internal logistics, and a hierarchical decomposition scheme is needed to handle sub-problems like task assignment, pickup and delivery dispatching [6], load selection, traffic control and general vehicle management [2]. The crux of the problem is to find a right combination and timing of these decisions [7] without compromising much the ideal logistics service. On the other hand, spatial decomposition, i.e., the segmentation of the material flow system into zones [8] remained still an efficient approach to tackle issues of congestion and collision avoidance on the shop-floor [2,9]. Network flow models are generic points of departure of state-of-the-art of fleet management methods, even on the level of dispatching [10]. However, new insight can be gained by analysing these models by means of the conceptual apparatus of network science: this way patterns of flow anomalies [11], or locus and boundaries of autonomous subsystems [12] can be detected.
The goal of this paper is to show that recent notions and methods of network science on modularity [13] can be well applied to detect the hidden structure of the material flow problem on the shop-floor. Once uncovered, this structuring can be exploited for balancing the expected load of vehicles and dispatching them dynamically. The paper presents a novel workflow for managing AGV fleets and the merits of its application on a series of comparative simulation studies taken from industrial experience.
Section snippets
Specification of the problems
The system under study consists of a set of machines and buffers as active material processing and passive storage stations, respectively, and an AGV fleet that transports the items among them in an automated way. Items are considered general container units of standard size, capable of holding any kind of input/output material of production. The AGVs are identical and can carry multiple items up to their maximal capacity. The flow of materials is determined by the routing, which defines the
Network analysis and load balancing
The initial problem is represented in terms of a material flow network (simply referred to as network) which combines main properties of the layout, the routings and the workload of the production system. Nodes of this network are the stations, whereas directed and weighted edges stand for the movement of items. The weight of any edge is (1) proportional to the number of items (to be) transported from the pickup to the destination node over a given period, and is (2) inversely proportional to
Dispatching
In the phase of dispatching, tasks are assigned to vehicles within their zones, assuming a saturated system where machines continuously trigger tasks. Each AGV maintains its own list of assigned tasks and their execution sequence is determined by the dispatcher. The proposed distance- and time-based dispatching (DTB) approach basically follows the pickup task-first policy [7], however, it considers not only the pickup, but also the delivery locations of items, as well as the current and future
Experimental results
The effectiveness of the complete workflow is demonstrated here via experimental results, taken from a large-scale industrial case study. The discrete-event simulation model (with a 95% validated accuracy) of the real system was used as a testbed of the experiments. The system consists of more than 150 machines and 20 AGVs with the capacity for transporting at most six items. Assuming a saturated system throughout the experiments, the main objective was to maximize machine utilizations by
Conclusions
The new AGV fleet management workflow benefits from the analysis of the overall material flow network: it reduces problem complexity in a hierarchical decomposition scheme into fleet planning and dispatching levels, and contributes to a spatial decomposition of the problem into zones. The proposed modularity-based clustering detects zones of stations with strong material flow dependencies, without the need of predefining their expected numbers. In this way, the adaptiveness of the overall
Acknowledgement
This research has been supported by the European H2020 EPIC grant under No. 739592 and the GINOP-2.3.2-15-2016-00002 grant of Hungary.
Declaration of Interests
None.
References (17)
- et al.
Versatile Autonomous Transportation Vehicle for Highly Flexible Use in Industrial Applications
CIRP Annals
(2012) - et al.
Automated Guided Vehicle Systems, State-Of-The-Art Control Algorithms and Techniques
Journal of Manufacturing Systems
(2020) Survey of Research in the Design and Control of Automated Guided Vehicle Systems
European Journal of Operational Research
(2006)- et al.
Innovative Control of Assembly Systems and Lines
CIRP Annals
(2017) - et al.
Automated Guided Vehicles Fleet Match-Up Scheduling with Production Flow Constraints
Engineering Applications of Artificial Intelligence
(2014) - et al.
A Multiple-Attribute Method for Concurrently Solving the Pickup-Dispatching Problem and the Load-Selection Problem of Multiple-Load AGVs
Journal of Manufacturing Systems
(2012) - et al.
Computational Intelligence in Control of AGV Multimodal Systems
IFAC-Papersonline
(2018) - et al.
Anomaly Detection in Shop Floor Material Flow: A Network Theory Approach
CIRP Annals
(2013)
Cited by (3)
From ethics to standards – A path via responsible AI to cyber-physical production systems
2022, Annual Reviews in ControlAdaptive AGV fleet management in a dynamically changing production environment
2020, Procedia ManufacturingAutomatic Drones for Factory Inspection: The Role of Virtual Simulation
2021, IFIP Advances in Information and Communication Technology