Method for creating a non-linear, stationary or dynamic model of a control variable of a machine

A modeling method for a control variable of a machine, by establishing a non-linear, stationary or dynamic model over an entire spectrum of all operating points of the machine, whereby the control variable is dependent on a set of system parameters, comprising the steps of:

• a) determining by way of measurement values of the control variable for a group of operating points out of the entire spectrum of operating points of the machine,
• b) determining an output quantity at the group of operating points using a partial model function modeling the machine behavior,
• c) adding at any operating point the output quantities of each partial model function in a weighted fashion to an associated weighting function to arrive at a total output quantity for the control variable of the engine for the respective operating point,
• d) determining for all operating points with the control variable determined by measurement a difference between total output quantity and value of the control value arrived at by measurement,
• e) choosing a further model function and using it with a further associated weighting function in such areas of operating points where an absolute value of this difference is above a preset value, such that an absolute value of the difference remains below the preset value, and
• f) repeating the steps of determining the difference between total output quantity of associated partial model functions and a value of the control variable of the engine at this operating point, that is evaluated by measurement, and of applying of a further model function and further weighting function as many times as needed until a statistically evaluated prediction quality of the model over the entire spectrum of all operating points of the machine has reached a desired value.