At the start of the project, problems in enterprise production scheduling often arise, such as unclear division of tasks, inflexible use of resources, and inefficient connections between processes; thus, lean goals are not met. Therefore, a model for lean enterprise scheduling pre-management based on optimisation algorithms will be developed. A work decomposition structure is introduced in the task parsing layer of the model, graphical modelling is used for order, process and equipment information, and an optimisation algorithm combining constrained programming and heuristic search determines the resource loading scheme. In the scheduling core layer, by strengthening the hybrid framework of genetic algorithm and tabu search and adding processing time, equipment status and work-in-progress stock, dynamic optimisation and batch rearrangement of production tasks have been achieved. A scheduling instruction caching mechanism based on a rule engine is used at the execution collaboration layer to complete data connection with the MES system and achieve synchronous updates of task flows and device signals. Based on six months of real work order data from a certain equipment manufacturing company, a total of 5,172 task records and 68 key devices were obtained for the experiment. Based on the above data, the shortened Makespan index by about 11.8% ±0.6, raised the resource utilisation rate to around 93.7% ±0.4, and increased the order fulfillment rate by about 9.6% ±0.5. The innovation of this study is as follows: propose an optimisation framework that integrates multiple heuristic operators; build a pre-model method for the link between task diagrams and resource diagrams; and construct an integrated management closed loop of scheduling and execution to provide a scalable technical path for lean production.
