The goal of modeling is the improvement of adjustment, optimization, conversion and control of shortened process chains. The behavior of a shortened process chain is emulated by various modeling methods. Using these methods, adjustment and optimization can be carried out by simulation so that many expensive experiments are no longer necessary. Each shortened process chain has its own properties. Therefore, it is impossible to use a single modeling method. For each shortened process chain, the most appropriate modeling method must be found. Artificial intelligence methods are particularly suitable for this purpose. Using artificial neural networks or knowledge-based systems, a shortened process chain can be modeled without knowledge of the exact correlations between the input and the output parameters. Certain topologies of neural networks can be compiled automatically and the neural networks are able to learn nearly all correlations. Knowledge-based systems can store quantitative and qualitative correlations and can draw conclusions from the output to the input parameters. In many cases, various modeling methods have to be combined to form a hybrid model. Using this model, the user can save a lot of money and time due to the reduced number of expensive experiments.
展开▼
