Closed-loop model parameter identification techniques for industrial model-based process controllers

An apparatus comprising:

• at least one memory configured to store closed-loop data associated with operation of an industrial process controller that is configured to control at least part of an industrial process using at least one model;
• and
• at least one processing device configured to:
• generate at least one noise model using a first model identification process associated with the industrial process controller using at least some of the closed-loop data, the first model identification process comprising a high-order autoregressive with exogenous terms (ARX) model identification process;
• filter the closed-loop data based on the at least one noise model, wherein the filter chosen is an estimation based on the generated at least one noise model is defined as:
• s(t,{circumflex over (η)}N)=A(q,{circumflex over (η)}N)s(t);
• wherein s(t) are the signals to be filtered, A(q) the ARX model and {circumflex over (η)}N represents estimates of model parameters;
• and
• generate one or more model parameters for the industrial process controller using the filtered closed-loop data, wherein generating the one or more model parameters comprises performing a second model identification process comprising an output-error (OE) model identification process.

A non-transitory computer readable medium containing instructions that, when executed by at least one processing device, cause the at least one processing device to:

• obtain closed-loop data associated with operation of an industrial process controller, the industrial process controller configured to control at least part of an industrial process using at least one model;
• generate at least one noise model using a first model identification process associated with the industrial process controller using at least some of the closed-loop data, the first model identification process comprising a high-order autoregressive with exogenous terms (ARX) model identification process;
• filter the closed-loop data based on the at least one noise model, wherein the filter chosen is an estimation based on the generated at least one noise model is defined as:
• s(t,{circumflex over (η)}N)=A(q,{circumflex over (η)}N)s(t);
• wherein s(t) are the signals to be filtered, A(q) the ARX model and {circumflex over (η)}N represents estimates of model parameters;
• and
• generate one or more model parameters for the industrial process controller using the filtered closed-loop data, wherein generating the one or more model parameters comprises performing a second model identification process comprising an output-error (OE) model identification process.