Time Lag Aware Deep Reinforcement Learning for Flexible Job Shop Scheduling in PPVC Module Factories

Jul 16, 2026

Prefabricated prefinished volumetric construction (PPVC) shifts most building work into module factories operating as flexible job shops. Long post-operation time-lags from concrete curing, watertightness ponding tests, and paint drying block modules while workstations remain available. In benchmark instances based on an official national guidebook, modeling lags increases even optimal makespans by about 67% on average, and lag-blind repair performs worse than all dispatching rules. A dual-attention deep RL solver is adapted via three ablations for lag-aware decision-making, and a released benchmark generator supports learned scheduling that replans within seconds after disruptions.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages10

Reading Time25 min

Abstract

Index Terms

I. Introduction

Prefabricated Prefinished Volumetric Construction (PPVC)

Problem Difference and Time-Lag Impact

Motivation and Need for Fast Lag-Aware Replanning

Time-Lag-Aware Deep Reinforcement Learning for Flexible Job-Shop Scheduling in PPVC Module Factories Ziheng Zhang, Student Member, IEEE, and Wei Zhang, Senior Member, IEEE arXiv :2607 . 1 1725v 1 [ cs .LG] 13 Jul 2026 Abstract—Prefabricated prefinished volumetric construction moves most building work into module factories, whose production floor operates as a flexible job shop. One complication is decisive: long post-operation time-lags caused by concrete curing, watertightness ponding tests, and paint drying, during which a module is blocked while its workstation stays free. On benchmark instances grounded in an official national prefabrication guidebook, these lags inflate even the optimal reference makespan by about 67% on average, and ignoring them at decision time, then repairing to feasibility, is worse than every dispatching rule. We adapt a state-of-the-art dual-attention deep reinforcement learning solver through three minimally invasive, individually ablatable extensions: lag-aware dynamics with an admissible reward bound, two anticipatory lag feature channels, and livenessmasked operation- and station-type embeddings. With every extension disabled the implementation reproduces the original solver exactly, so all gains are attributable to the adaptations. We release a public, guidebook-grounded benchmark generator. On held-out instances the learned policy is the strongest solverfree scheduler: it reaches within about 4% of a constraintprogramming reference and beats every dispatching rule anda genetic-algorithm metaheuristic, with its advantage widening under capacity contention, and a single size-mixed policy carries this lead across the trained range of factory sizes. It needs nosolver, model, or license in the loop and re-plans within seconds of a disruption; where an exact solver can be deployed, that solver remains the quality ceiling, a boundary we map explicitly. Index Terms—Deep reinforcement learning, flexible job-shop scheduling, prefabricated prefinished volumetric construction (PPVC), reactive rescheduling, time-lags. I. INTRODUCTION PREFABRICATED Prefinished Volumetric Construction (PPVC) relocates structural work, mechanical, electrical, and plumbing (MEP) installation, and interior finishing from the construction site into a factory that mass-produces fully fitted three-dimensional modules [1], [2] . Singapore’s Building and Construction Authority (BCA), for instance, codifies PPVC production requirements in an official guidebook [3] and requires PPVC on selected government land-sale sites, making module-factory throughput a binding constraint on national housing programs. The factory floor is a flexible job shop: each module visits a sequence of specialized stations Z. Zhang (Research Fellow) and W. Zhang (Associate Professor) are with the Singapore Institute of Technology, Singapore (e-mail: zi[heng.zhang@singaporetech.edu.sg](heng.zhang@singaporetech.edu.sg); [wei.zhang@singaporetech.edu.sg](wei.zhang@singaporetech.edu.sg)). (Corresponding author: Wei Zhang.) Manuscript submitted July 2026 . This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible. TABLE I MAKESPAN INFLATION WHEN TIME-LAGS ARE MODELED (5-MODULE MIXED INSTANCE , DEFAULT FACTORY; FIFO: FIRST-IN-FIRST-OUT) . | Scheduler | Lag-Free (h) | Lag-Aware (h) | Inflation | | --- | --- | --- | --- | | FIFO dispatching | 113.0 | 206.0 | +82% | | CP-SAT optimal | 100.0 | 198.0 | +98% | (mold preparation, casting or welding, MEP fit-out, tiling, painting, quality gates), and each visit may be served by any station of a compatible type. What separates PPVC factories from the classical flexible job-shop scheduling problem (FJSP) is the prevalence of postoperation time-lags. After concrete is poured, the module must cure for 24–48 h; after waterproofing, wet modules undergo a ponding test of 24–48 h; each paint coat

FAQs

Why are time-lags critical in PPVC module factory scheduling?

Modules are blocked during concrete curing, ponding tests, and paint drying, while the producing stations can immediately serve other modules. These lags form the dominant temporal structure rather than a minor disturbance.

What happens to makespan when time-lags are modeled versus ignored?

On lag-aware instances, the optimal makespan increases substantially (e.g., a 5-module example rises from 100.0 h to 198.0 h, and the broader mean inflation is reported as 67.2%). Ignoring lags at decision time and repairing afterward yields worse performance than dispatching rules.

How does the proposed method achieve lag-aware and fast rescheduling?

It adapts a dual-attention deep reinforcement learning solver with three minimally invasive extensions: lag-aware dynamics with an admissible reward bound, two anticipatory lag feature channels, and liveness-masked embeddings. The learned policy replans within seconds after disruptions and does not require solver-in-the-loop execution.

Jul 16, 2026

Upload by Valentina