Research on scheduling optimization of ship plate processing workshop based on improved NSGA-II algorithm

Why shipyards need smarter schedules

Modern shipyards handle thousands of heavy steel plates that must be marked, cut, and moved in just the right order. Any small disruption—like a broken cutting machine or a rush order—can ripple through the workshop, wasting energy, overworking some machines, and threatening delivery dates. This paper presents a new way to automatically reorganize work in a ship plate processing workshop when such disruptions occur, using a refined evolutionary algorithm to keep production fast, reliable, and efficient.

Keeping production steady when things go wrong

Shipbuilding is a complex, stop‑and‑go kind of manufacturing. Plates vary in size and shape, and different machines share the workload. Today, when something unexpected happens, many shipyards still rely on experienced staff to reshuffle the plan by hand. That takes time and often leads to uneven machine use and higher costs. The authors focus on a key question: when the shop floor is hit by events like machine failures, rework, or late materials, how can a computer quickly generate a new plan that finishes on time, keeps energy use low, and avoids overloading any single machine?

Turning the workshop into a digital twin

To tackle this, the researchers first turn the ship plate workshop into a detailed digital model. They build a three‑dimensional layout of machines and material flows using engineering software, and link it with an Internet of Things (IoT) data platform that collects real‑time information from cutting tables, cranes, and other equipment. This creates a kind of digital twin of the workshop: a virtual environment that mirrors what is happening on the floor. Production data flow into a scheduling system, which uses optimization algorithms to propose an initial work plan. That plan is then tested in simulation to check whether it respects delivery deadlines and uses machines reasonably before being sent back to control the real workshop.

Balancing time, cost, and machine workload

The heart of the study is a mathematical description of how plates move through the workshop. Each plate passes through several steps on different machines, and the plan must respect the order of operations, the capacity of each machine, and a promised delivery time. The authors define three goals at once: shorten the overall completion time, reduce the total energy used during processing and standby, and avoid long periods where machines are either idle or overloaded. This kind of multi‑goal problem has no single perfect answer. Instead, it produces a set of trade‑offsfor example, finishing slightly earlier at the cost of higher energy use. The goal of the algorithm is to map out these trade‑offs so planners can choose a schedule that best matches their priorities.

Teaching an algorithm to adapt like an expert

To search through the enormous space of possible schedules, the authors improve on a popular evolutionary method called NSGA‑II, which works by evolving a population of candidate plans over many generations. Traditional versions use fixed settings for how often plans are mixed and randomly changed, and they preserve the best plans in a simple way. This can cause the search to get stuck too early. Here, the probabilities of mixing and mutation adapt automatically as the search progresses, encouraging wide exploration at the start and more careful refinement later. At the same time, a new elite selection rule, inspired by simulated annealing, controls how many of the best plans are kept from each generation. This helps maintain variety among promising schedules so that the algorithm does not converge too quickly on a sub‑optimal solution.

Proving the method in tests and a real shipyard

The improved approach is tested in two ways. First, it is run on a suite of standard scheduling benchmarks widely used by researchers. Across most of these tests, it finds more diverse and higher‑quality trade‑off solutions than both the original NSGA‑II and a newer variant called NSGA‑III. Second, the team applies it to a real production order involving 16 plates and seven machines in a shipyard, then introduces realistic disruptions: rush rework jobs and a major machine breakdown. In each case, the system first tries a simple right‑shift of affected tasks; if that would miss the delivery date, it triggers a full rescheduling using the improved algorithm. Compared with traditional strategies, the new method delivers shorter completion times, lower or similar energy use, and better balanced machine workloads, while still computing plans fast enough for practical use.

What this means for shipbuilding

For non‑specialists, the key message is that ship plate workshops can now respond to surprises in a more automatic and reliable way. By combining a live data stream from the factory, a realistic digital model, and a smarter evolutionary algorithm, the method keeps production on schedule with less manual firefighting. In the long run, such dynamic scheduling could help shipyards reduce delays, save energy, and make better use of expensive equipmenta concrete step toward more intelligent, resilient manufacturing.

Citation: Dong, L., Liu, J., Gu, S. et al. Research on scheduling optimization of ship plate processing workshop based on improved NSGA-II algorithm. Sci Rep 16, 5549 (2026). https://doi.org/10.1038/s41598-026-35278-y

Keywords: shipbuilding, production scheduling, genetic algorithm, smart manufacturing, dynamic optimization