Engineering Tool and Method for Parameterizing a Model-Based Predictive Controller

An engineering tool for parameterizing a model-based predictive controller for controlling a engineering process, the engineering tool comprising:

• a quality determining module for determining a quality of a model that the model-based predictive controller contains for response of the engineering process to be controlled;
• and
• a weighting factor module for automatically determining weighting factors used to weight manipulated variable changes in a quality criterion, the model-based predictive controller being used to minimize said quality criterion, and the determination of the weighting factors being dependent on a respectively determined quality of the model;
• wherein the quality determining module for determining the quality of the model is configured to utilize measurement data to ascertain errors in a plurality of model parameters as model errors, and
• wherein the weighting factor module for automatically determining weighting factors is configured to determine the weighting factors such that the manipulated variable changes in a model-based predictive controller designed taking into account model errors are essentially the same as those in a predictive controller designed based on an assumption of an error-free model.

A computer implemented method for parameterizing a model-based predictive controller for controlling an engineering process, the method comprising the steps of:

• determining, in a processor of the computer, a quality of a model that the model-based predictive controller contains for a response of the engineering process to be controlled;
• automatically determining weighting factors for providing weighted manipulated variable changes in a quality criterion;
• minimizing, by the model-based predictive controller, the weighted manipulated variable, determination of the weighting factors being dependent on a respectively determined quality of the model;
• ascertaining, based on measurement data, errors in a plurality of model parameters as model errors;
• and
• determining the weighting factors such that the manipulated variable changes in a model-based predictive controller designed taking into account the model errors are essentially the same as those in a predictive controller designed based on an assumption of an error-free model.