In this European funded project an Expert System in sizing was developed, which includes all the relevant informations about the complex sizing operation. The system allows to evaluate the optimum sizing conditions for best weaving. The system is able to learn and enlarge its knowledge by algorithms on the base of software utilising neural networks.
It includes the relevant knowledge, which was available in four first class textile mills, a producer of textile machines, a textile research institute and several size producers. In comprehensive tests the knowledge was extended to the influence of viscosity to penetration and weavability under consideration of the new prewetting sizing system. For the practical tests in the mills a control system for size add-on was used, to give the necessary security in sizing and weaving efficiency. Relations between the necessary size add-on to the fabric construction and weaving parameters are considered. The output of the system are the settings of the sizing machine (elongation and tension in all parts of the sizing machine, squeezing pressure, production speed, drying temperature). These parameters are calculated on base of several size receipts, which gives an optimum in costs or a minimum in environmental impacts. They are found by evolutionary strategies.