Neural net implementation for assigning a product to a production line

Summary form only given. Assigning a new product to a production line is one of many problems whose solution is very complex and is approached by either exact mathematical programming or quick heuristics. The solution proposed imitates a foreman´s decision when he faces a real problem. A perceptron-type neural network is developed whose input parameters are a function of planning data, real-time status, and local expertise. Its output is the foreman´s decision. The robustness of this approach is demonstrated by a case study.<>