Production Scheduling Optimization: 66% Adherence Increase

In multi-variety, small-batch production of around 30,000 products, numerous variables and different equipment/processes are involved. Skilled operators often struggle to manage these variables due to the complexity and dynamic nature of the production process, causing bottlenecks and idle equipment. Effective production scheduling is needed to improve delivery adherence and optimize output.

An AI-based simulator was developed to automate the batching and evaluation of work processes. Using this simulator that mimics actual production schedules, reinforcement learning agents derived optimal scenarios to minimize bottlenecks and downtime. An AI operational system (MLOps) was built to validate data and rapidly improve models based on real-world data.

Schedule adherence increased from 19% to 85% in multi-variety, small-batch production. Delivery delays were prevented, and productivity boosted by quickly optimizing scheduling to handle unexpected changes like equipment breakdowns and order modifications.