基于WOA-IGWO-LSTM的作业车间实时调度  

Real-Time Scheduling of Job Shop Based on WOA-IGWO-LSTM

作　　者：郑华丽 魏光艳 孙东 王明君 叶春明[2] ZHENG Huali;WEI Guangyan;SUN Dong;WANG Mingjun;YE Chunming(China Tobacco Zhejiang Industrial Co.,Ltd.,Hangzhou Zhejiang 310008,China;Business School,University of Shanghai for Science&Technology,Shanghai 200093,China)

机构地区：[1]浙江中烟工业有限公司,浙江杭州310008 [2]上海理工大学管理学院,上海200093

出　　处：《机床与液压》2025年第2期54-63,共10页Machine Tool & Hydraulics

基　　金：上海市哲学社会科学一般项目(2022BGL010)。

摘　　要：针对作业车间实时调度问题,基于长短期记忆(LSTM)神经网络,提出WOA-IGWO-LSTM算法。根据调度问题和算法设计三元样本数据结构,以性能指标和生产系统状态属性作为输入特征,输出当前决策点的最佳调度规则。利用鲸鱼优化算法(WOA)对输入特征进行降维,以提高模型泛化能力和准确性。引入非线性收敛因子设计一种改进灰狼算法(IGWO)用于调节LSTM参数,提高算法实用性。最后,通过对比试验验证了WOA、IGWO以及WOA-IGWO-LSTM的有效性,并利用工业案例数据验证了WOA-IGWO-LSTM对于解决作业车间实时调度问题的有效性和可行性。Aiming at the real-time job shop scheduling problem,a WOA-IGWO-LSTM algorithm was proposed based on long short term memory(LSTM)neural network.According to the scheduling problem and algorithm,a ternary sample data structure was designed,and the performance index and production system state attributes were used as input features to output the optimal scheduling rules for the current decision point.The whale optimization algorithm(WOA)was used to reduce the dimensionality of the input features to improve the generalization ability and accuracy of the model.An improved grey wolf optimization(IGWO)algorithm was designed by introducing the nonlinear convergence factor to adjust the LSTM parameters and improve the practicability of the algorithm.Finally,the effectiveness of WOA,IGWO and WOA-IGWO-LSTM is verified by comparative experiments,and the effectiveness and feasibility of WOA-IGWO-LSTM for solving real-time job shop scheduling problem are verified by using industrial case data.

关 键 词：长短期记忆(LSTM)神经网络 鲸鱼优化算法(WOA) 改进灰狼算法 作业车间实时调度 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]