AI-based Predictive Dynamic Production Planner

The Experiment, conducted by Hohner Automaticos, envisaged an AI-based Predictive Dynamic Production Planner in an SME. Currently, the biggest productivity losses occur in product change tasks (more or less 30%-40% of the total); these setbacks can be reduced by a better production planning & scheduling to reduce machine downtimes, which is especially convenient when high levels of product customization are requested. Similarly, the incidents that occur in plant day-to-day (for example, breakdowns, supplier delays, non-conforming materials, noncompliance with standards, poorly manufacturing product, among others) lead to over costs and delays in deliveries: having connected information in real time and with the ability to analyse it, to propose immediate corrective actions would allow to redirect these conflict situations and modify the priorities avoiding delivery delays to the customer. This approach is simple but cumbersome and does not readily adapt to changes in demand and products, assets capacity and availability -both human and material- due to assets and capacity are planned separately, and the assets or capacity constraints are not considered, leading to infeasible plans. Thus, large direct and indirect costs, excess inventories and incidents in customer service are generated.

The experiment has introduced a planning and scheduling module for production process. The module includes two smart tools, one for production orders sequencing based on available assets, labour and plant capacity, and another for demand forecasting. The planning and scheduling module is linked with the legacy systems in order to support the production manager and inventory manager in the day to day on the shop floor. The experiment aimed to improve the efficiency of production and reliability of supplies due to an optimization of the use of manufacturing orders, stock knowledge, human capabilities, assets availability, among others. Moreover, the dynamic planning and scheduling helps to: (i) know the status and priority of each works order on the shop floor; (ii) know each operation in the manufacturing process; (iii) know which assets or resources are required for each works order; (iv) know when the assets or resources are required and when they will be available; and (v) know the timings and delays between operations for more realistic production lead times.