You are here

Progress at Sztaki

In the 2nd Demonstrator led by our industrial partner Knorr-Bremse, a novel method is elaborated that focuses on increasing the robustness of the plant-level production planning. The task considers the changes and possible disturbances arising on the shop-floor level in the higher level production planning. The planning is improved by combining statistical learning, mathematical optimization and discrete-event simulation.
The new planning methods are being validated and evaluated by applying them on a pilot flexible assembly line which is responsible for the production of several different product variants. Important factors in the production planning are the variable processing times, rework rates and different capacity requirements. The goal of the method is to increase the robustness of the production plans and optimize the resource allocation in the short and medium terms.
The overall methodology relies on the detailed simulation model of the line that represents the stochastic factors like variable processing times, availability of the resources and rework rates. In order to include them in the mathematical optimization models, statistical learning methods were applied that predict the important planning parameters. According to the first results, it is possible to balance the workload of the operators, increase the robustness of the plan and therefore reduce the extra costs e.g. the amount of backlogs.
On the basis of the first promising results, in the second part of RobustPlaNet project both, the developed methods and the simulation model will be integrated, tested and validated into the RobustPlaNet Cockpit on industrial site.

Powered by Drupal