💻 computer science

Adaptive Mutation Driven Harris Hawk Optimization Algorithm in Hybrid Flow Workshop Scheduling with Robotic Arm

The paper proposes an Adaptive Mutation Driven Harris Hawk Optimization (AMDHHO) algorithm that integrates Levy flight and dynamic mutation strategies to effectively solve hybrid flow shop scheduling problems with robotic arms, demonstrating significant reductions in maximum completion time and average waiting time compared to genetic algorithms and original scheduling plans.

Original authors: Bin Li

Published 2026-07-07

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Bin Li

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a busy factory floor as a giant, high-stakes game of musical chairs, but instead of people, you have robotic arms moving workpieces (like car parts) between different workstations. The goal is to get every single part finished as quickly as possible without any robot bumping into another or any machine sitting idle.

This is the problem the paper tackles: How do you schedule a factory with robotic arms so everything runs perfectly fast?

Here is the simple breakdown of the author's solution:

1. The Problem: A Chaotic Dance

In a "Hybrid Flow Workshop," parts go through several stages (like turning, milling, drilling). At each stage, there are multiple machines, and robotic arms are needed to carry the parts from one machine to the next.

The Challenge: If you schedule this poorly, robots wait around, machines sit idle, and the whole factory slows down.
The Old Way: Humans try to plan this, or computers use older "smart" algorithms (like Genetic Algorithms). But these often get stuck in a rut, finding a "good enough" solution but missing the perfect one, or they take too long to figure it out.

2. The Inspiration: The Harris Hawk

The author looked to nature for a better way. Harris Hawks are birds that hunt in packs. They don't just fly randomly; they use a clever strategy:

The Chase: They scout the area (exploration).
The Ambush: Once they spot prey, they surround it and attack from different angles (exploitation).
The Twist: They adapt their strategy based on how tired the prey is.

The author turned this bird behavior into a computer algorithm called HHO (Harris Hawk Optimization).

3. The Upgrade: The "Adaptive Mutation" Boost

The standard bird algorithm is good, but the author realized it needed a "turbocharger" for complex factory problems. They created a new version called AMDHHO (Adaptive Mutation Driven Harris Hawk Optimization).

Think of the algorithm as a team of bird-simulators trying to find the best schedule. To make them smarter, the author added two special tricks:

Levy Flight (The "Magic Leap"): Sometimes, instead of taking small, logical steps, the birds take a giant, random leap to a completely different part of the map. This helps them escape "local traps" (bad schedules that look good but aren't the best) and find hidden, better solutions.
Adaptive Mutation (The "Dynamic Shuffle"): Imagine the birds are trying to solve a puzzle. Early on, they are wild and try crazy new combinations (high mutation). As they get closer to the finish line, they become more careful and fine-tune their moves (low mutation). The algorithm automatically knows when to be wild and when to be precise.

4. How It Works in the Factory

The algorithm translates the factory problem into a "code" (a chromosome) that looks like a to-do list:

Who goes first? (The order of parts).
Which machine? (Which of the parallel machines gets the part).
Which robot? (Which arm carries the part).

The algorithm runs thousands of simulations in seconds, shuffling these lists using its "bird tactics" to find the schedule that finishes everything in the shortest time.

5. The Results: Faster and Smarter

The author tested this new algorithm in two ways:

The Test Track (Standard Examples): They ran it against standard factory problems with 10 to 50 parts.
- The Result: AMDHHO was the clear winner. It finished the "race" faster than the old Genetic Algorithm (GA) and the standard Hawk algorithm. For the biggest test, it saved about 11.7% of the total time compared to the old methods. It was also more consistent, rarely making mistakes (low standard deviation).
The Real World (A Factory in Suzhou, China): They applied it to a real factory making car and airplane parts with 20 parts and 3 robots.
- The Result: The new schedule cut the total time from 768 minutes down to 683 minutes. That's 85 minutes saved per batch!
- Bonus: It also reduced the time parts sat waiting (waiting time) by nearly 40%.
- Speed: If a machine broke down, the algorithm could recalculate a new plan in just 9.6 seconds.

Summary

The paper claims that by copying how Harris Hawks hunt and adding a "dynamic shuffle" mechanism, the author created a computer program that schedules factory robots much better than current methods. It gets the job done faster, keeps the robots busier, and handles real-world chaos better than the old ways of doing things.

Technical Summary: Adaptive Mutation Driven Harris Hawk Optimization Algorithm in Hybrid Flow Workshop Scheduling with Robotic Arm

Problem Statement
The paper addresses the Hybrid Flow Shop Scheduling Problem with Loading/Unloading Machines (HFSP-LUM), a complex combinatorial optimization challenge prevalent in intelligent manufacturing. This specific variant involves scheduling $n$ workpieces across $m$ processes, where each process contains parallel identical equipment, and crucially, incorporates robotic arms responsible for material handling between adjacent stages. The problem is characterized by high dimensionality, strong constraints (e.g., non-preemptive processing, single-task robotic arms, process sequences), and the need to minimize the maximum completion time ( $C_{max}$ ). The authors note that traditional manual scheduling and existing heuristic algorithms often struggle with dynamic production demands, premature convergence, and insufficient global search capabilities in such high-dimensional, constrained environments.

Methodology
To address these challenges, the authors propose the Adaptive Mutation Driven Harris Hawk Optimization Algorithm (AMDHHO). The methodology integrates a mathematical model of the HFSP-LUM with an enhanced meta-heuristic solver.

Mathematical Modeling:
- A mixed-integer programming model is constructed with the objective of minimizing $C_{max}$ .
- The model includes constraints for process sequences, machine allocation (one workpiece per machine at a time), robotic arm exclusivity (one transport at a time), and temporal dependencies (start times must respect transport and processing completion times).
- Decision variables include workpiece processing sequences, machine assignments, and robotic arm task allocations.
Algorithm Design (AMDHHO):
- Base Framework: The algorithm is built upon the standard Harris Hawk Optimization (HHO) framework, which simulates the "ambush encirclement" hunting tactics of Harris hawks, alternating between exploration (random patrolling) and exploitation (tracking and attacking prey).
- Hybrid Encoding: A three-dimensional hybrid encoding structure is used to represent solutions:
  - Processing Sequence Vector (PSV): Determines workpiece priority.
  - Machine Assignment Vector (MAV): Assigns specific parallel machines to workpieces.
  - Robot Arm Assignment Vector (RAV): Allocates robotic arms for inter-stage transport.
- Key Improvements:
  - Levy Flight Disturbance: Integrated into the search mechanism to enhance global exploration and prevent premature convergence.
  - Adaptive Mutation Probability: A dynamic strategy adjusts the mutation probability based on the iteration count and population energy state, balancing global search in early stages with local development in later stages.
  - Multi-method Crossover: Utilizes Partially Mapped Crossover (PMX) for sequence optimization and Uniform Crossover for machine/robot allocation, with adaptive probability control.
  - Decoding Strategy: A step-by-step decoding mechanism translates the 3D vectors into executable schedules, calculating start/completion times while strictly adhering to resource and temporal constraints.

Key Contributions

Algorithmic Innovation: The development of AMDHHO, which specifically targets the HFSP-LUM by integrating adaptive mutation and Levy flight mechanisms into the HHO framework.
Comprehensive Modeling: The construction of a rigorous mathematical model that explicitly accounts for robotic arm constraints and transport times, which are often simplified or ignored in standard flow shop models.
Hybrid Encoding & Decoding: The proposal of a specific encoding/decoding scheme tailored to the multidimensional decision variables of robotic-assisted hybrid flow shops.
Empirical Validation: Extensive testing on both standard benchmark instances (Taillard-based) and a real-world industrial case study involving 20 workpieces, 11 machines, and 3 robotic arms.

Experimental Results
The authors conducted comparative experiments against the standard HHO, Mayfly Algorithm (MA), Genetic Algorithm (GA), and Beetle Optimization Algorithm (DO) across five scales of problem instances (ranging from 10 to 50 workpieces).

Standard Benchmarks:
- In the largest scale test (50 workpieces, 5 stages, 5 machines), AMDHHO achieved a $C_{max}$ of 492.6 minutes.
- This represents an improvement of approximately 11.7% over the Genetic Algorithm (557.8 minutes) and 5.0% over the standard HHO.
- AMDHHO demonstrated superior stability, with a standard deviation of 3.27, significantly lower than the GA's 7.92.
- Convergence speed was notably faster; AMDHHO converged in 57 iterations, whereas GA required 127 iterations for the same problem scale.
Ablation Studies:
- Experiments confirmed that the combination of adaptive mutation and Levy flight was the primary driver of performance, improving optimal solutions by ~13% and average solutions by ~15% compared to the standard HHO.
- The adaptive mutation mechanism significantly reduced solution volatility (standard deviation reduction of 42.6–45.2%).
Real-World Application (Suzhou Precision Parts Manufacturing):
- Applied to a production line with 20 workpieces and 3 robotic arms.
- The AMDHHO reduced the maximum completion time from 768 minutes (original manual plan) to 683 minutes, a reduction of 85 minutes (11.1%).
- Average waiting time decreased by 39.8%, and resource utilization increased to 82.7%.
- In dynamic scenarios simulating equipment failures, the algorithm generated adjusted schedules in approximately 9.6 seconds, demonstrating real-time adaptability.

Significance and Claims
The paper claims that the AMDHHO algorithm provides an efficient and robust solution for complex workshop scheduling in the context of intelligent manufacturing. The authors assert that the algorithm successfully balances global exploration and local development, overcoming the limitations of traditional swarm intelligence algorithms in high-dimensional, strongly constrained environments. The results suggest that AMDHHO offers significant practical value for reducing production cycles and improving resource utilization in robotic-assisted hybrid flow workshops, with broad engineering application prospects. The study emphasizes that the algorithm's dynamic adaptive mechanisms make it particularly suitable for environments requiring rapid response to dynamic events such as equipment failures.