Method and device for conducting a process in a controlled system with at least one precomputed process parameter determined using a mathematical model having variable model parameters adjusted based on a network response of a neural network

A method for controlling a process with a controlling means and a computing means, the computing means having a mathematical model of the process, the mathematical model having at least one variable process model parameter, and with a neural network having variable network parameters, the method comprising steps of:

• a) supplying input variables to the computing means;
• b) precomputing, before each process run, at least one selected process parameter based on the input variables supplied to the computing means;
• c) presetting the at least one process parameter;
• d) measuring input values and process parameters during a process run, wherein the process run is controlled by the controlling means;
• e) supplying the measured input values and process parameters to the mathematical model;
• f) adaptively improving the precomputed process parameters based on the measured input values and process parameters, the step of adaptively improving the precomputed process parameters including sub-steps of:
• i) supplying input values that influence the at least one variable process model parameter to the neural network after the process run,
• ii) providing a value for the at least one variable process model parameter as a network response with the neural network,
• iii) comparing the precomputed at least one process parameter with the measured process parameters to form a deviation, and
• iv) adapting the network parameters of the neural network such that the deviation is reduced.