Design and implementation of a fully automated planner-scheduler constraint satisfaction problem

The idea of constraint programming is to solve problems by stating constraints (conditions, properties) which must be satisfied by the solution. This paper introduces a fully automated scenario for complex scheduling problems. There are two constraint satisfaction problems: the planner (determines which orders should be accepted) and the scheduler (determines the timetable for the products). The third main component of the architecture is an OPC UA server which uses the solutions of the scheduler in order to control the devices of the machine tools, on which the parts are manufactured. An important step has been the reduction of the solving time corresponding to the second CSP (the scheduler). Two important actions have been taken. First the model has been split into four distinct CSPs, one for each manufacturing stage. Thus locally optimum solutions are combined into a global solution which is comparable to the global optimum solution. Secondly, we have tested various search strategies and we have managed to reduce the solving time to less than half of the initial time.