This paper addresses the challenge of integrated optimization for storage space allocation and crane scheduling in automated storage and retrieval systems. The problem encompasses tasks such as assigning storage/retrieval requests, allocating storage spaces, and planning crane routes within each operation cycle. To tackle this, we introduce a multi-layer adaptive length coding method to effectively map the solution space to the problem space. Employing a coevolutionary framework, we decompose and process the integrated optimization problem, further optimize it with a hybrid genetic algorithm. Numerical experiments across a wide range of scenarios are conducted to evaluate the algorithm’s performance under varying request sizes and crane capacities. The introduction of the coevolutionary framework improves optimization by up to 14.78%, with an average improvement of 34.09% compared to the method currently used in the company. In addition, we introduce a novel optimization metric, termed potential energy consumption, designed to enhance system energy efficiency. Comparative analysis against metrics like makespan reveals the superiority of our proposed approach in terms of coverage and optimality, particularly in large-scale scenarios. The combined implementation of integrated optimization and the new evaluation metric leads to substantial energy cost savings for real-world automated storage and retrieval systems.

中文翻译：

---

在自动存储和检索系统中集成优化存储空间分配和起重机调度


本文解决了自动存储和检索系统中存储空间分配和起重机调度的集成优化挑战。该问题包括在每个操作周期内分配存储/检索请求、分配存储空间和规划起重机路线等任务。为了解决这个问题，我们引入了一种多层自适应长度编码方法，以有效地将解决方案空间映射到问题空间。采用协同进化框架，我们分解和处理集成优化问题，并使用混合遗传算法进一步优化它。在各种场景中进行了数值实验，以评估算法在不同请求大小和起重机容量下的性能。协同进化框架的引入将优化提高了 14.78%，与公司目前使用的方法相比，平均提高了 34.09%。此外，我们还引入了一种新的优化指标，称为潜在能耗，旨在提高系统能效。与 makespan 等指标的比较分析揭示了我们提出的方法在覆盖率和最优性方面的优越性，尤其是在大规模场景中。集成优化和新评估指标的结合为实际的自动存储和检索系统节省了大量能源成本。