[Paper Review] Integrating Genetic Algorithm, Tabu Search Approach for Job Shop Scheduling

R. Thamilselvan, P. Balasubramanie|ArXiv.org|Jun 27, 2009

Scheduling and Optimization AlgorithmsEngineering13 references16 citations

TL;DR

This paper proposes a hybrid metaheuristic combining Genetic Algorithms (GA) and Tabu Search (TS) to solve the NP-complete Job Shop Scheduling problem, aiming to minimize makespan. The approach uses GA for global exploration and TS for local intensification, achieving optimal or near-optimal solutions more efficiently than either method alone.

ABSTRACT

This paper presents a new algorithm based on integrating Genetic Algorithms and Tabu Search methods to solve the Job Shop Scheduling problem. The idea of the proposed algorithm is derived from Genetic Algorithms. Most of the scheduling problems require either exponential time or space to generate an optimal answer. Job Shop scheduling (JSS) is the general scheduling problem and it is a NP-complete problem, but it is difficult to find the optimal solution. This paper applies Genetic Algorithms and Tabu Search for Job Shop Scheduling problem and compares the results obtained by each. With the implementation of our approach the JSS problems reaches optimal solution and minimize the makespan.

Motivation & Objective

• To address the computational intractability of the NP-complete Job Shop Scheduling (JSS) problem.
• To improve solution quality and convergence speed by integrating the global search capability of Genetic Algorithms with the local intensification of Tabu Search.
• To minimize the makespan—the total time to complete all jobs—across a set of jobs and machines.
• To evaluate the performance of the hybrid approach against standalone GA and TS methods.
• To demonstrate the effectiveness of hybrid metaheuristics in solving complex scheduling problems.

Proposed method

• The algorithm uses a genetic representation of job sequences, where each chromosome encodes a schedule.
• Crossover and mutation operators are applied to generate new candidate solutions in the genetic algorithm phase.
• Tabu Search is applied iteratively to local neighborhoods of promising solutions to escape local optima.
• A tabu list prevents revisiting recently explored solutions, enhancing diversification.
• The hybrid framework alternates between GA-based exploration and TS-based exploitation to balance solution quality and search efficiency.
• Fitness evaluation is based on minimizing the makespan, the primary objective function.

Experimental results

Research questions

• RQ1 Can the integration of Genetic Algorithms and Tabu Search yield better makespan results than either method used independently?
• RQ2 How does the hybrid approach perform in terms of convergence speed and solution quality on standard JSS benchmarks?
• RQ3 To what extent does the tabu list mechanism improve the search process by preventing cycling and enhancing diversification?
• RQ4 Does the combination of global and local search strategies lead to faster identification of optimal or near-optimal schedules?
• RQ5 How robust is the hybrid algorithm across different job shop scheduling instances?

Key findings

• The hybrid GA-TS approach achieved optimal or near-optimal solutions for benchmark job shop scheduling instances.
• The integration significantly improved convergence speed compared to standalone Genetic Algorithms.
• The use of Tabu Search enhanced local search capability, reducing the likelihood of getting trapped in local optima.
• The proposed method outperformed both individual GA and TS in terms of solution quality and computational efficiency.
• The makespan was minimized effectively, demonstrating the effectiveness of the hybrid strategy.
• The results validate the synergy between global exploration (GA) and local intensification (TS) in solving complex scheduling problems.