Equipment maintenance is momentous for improving production efficiency, how to integrate maintenance into production to address uncertain problems has attracted considerable attention. This paper addresses a novel approach for integrating preventive maintenance (PM) into production planning of a complex manufacturing system based on availability and cost. The proposed approach relies on two phases: firstly, this study predicts required capacity of each machine through extreme learning machine algorithm. Based on analyzing historical data, the opportunistic periods are calculated for implementing PM tasks to have less impact on production and obtain the PM interval and duration. Secondly, this study obtains the scheduling planning and the least number of maintenance personnel through an improved ant colony optimization algorithm. Finally, the feasibility and benefits of this approach are investigated based on empirical study by using historical data from real manufacturing execution system and equipment maintenance system. Experimental results demonstrate the effectiveness of proposed approach, reduce personnel number while guarantee the maintenance tasks. Therefore, the proposed approach is beneficial to improve the company’s production efficiency.

设备维护对于提高生产效率非常重要，如何将维护整合到生产中以解决不确定的问题已经引起了广泛的关注。本文提出了一种基于可用性和成本将预防性维护（PM）集成到复杂制造系统的生产计划中的新颖方法。所提出的方法依赖于两个阶段：首先，本研究通过极限学习机算法预测每台机器的所需容量。在分析历史数据的基础上，计算机会周期以执行PM任务，以减少对生产的影响并获得PM间隔和持续时间。其次，本研究通过改进的蚁群算法获得调度计划和最少的维修人员。最后，通过实证研究，利用来自实际制造执行系统和设备维护系统的历史数据，研究了这种方法的可行性和优势。实验结果证明了该方法的有效性，在保证维护任务的同时减少了人员数量。因此，所提出的方法有利于提高公司的生产效率。