Real-time adaptation of system high fidelity model in feature space

A method to model an industrial asset, comprising:

• outputting, by an augmented system model that includes a system fidelity model of the industrial asset in series with a data driven model, a feature space version of augmented system model output calculated via execution of a global feature calculation, wherein the global feature calculation comprises calculating a local feature vector for each of multiple given components of the industrial asset, and combining the multiple local feature vectors into a global feature vector, wherein the feature space version of the augmented system model output is an n-dimensional representation of the augmented system model output and “n” is the number of measurable features in the augmented system model output;
• receiving, at a model adaptation element from a plurality of monitoring nodes, a series of current monitoring node values over time that represent a current operation of the industrial asset, wherein the industrial asset is a physical machine and the plurality of monitoring nodes are sensors connected to the industrial asset;
• calculating a feature space version of current operation of the industrial asset;
• comparing, by the model adaptation element, the feature space version of the augmented system model output with the feature space version of current operation of the industrial asset;
• and
• adapting parameters of the data driven model based on a result of said comparison.

A non-transitory, computer-readable medium storing instructions that, when executed by a computer processor, cause the computer processor to perform a method to model an industrial asset, the method comprising:

• outputting, by an augmented system model that includes a system fidelity model of the industrial asset in series with a data driven model, a feature space version of augmented system model output calculated via execution of a global feature calculation, wherein the global feature calculation comprises calculating a local feature vector for each of multiple given components of the industrial asset, and combining the multiple local feature vectors into a global feature vector, wherein the feature space version of the augmented system model output is an n-dimensional representation of the augmented system model output and “n” is the number of measurable features in the augmented system model output;
• receiving, at a model adaptation element from a plurality of monitoring nodes, a series of current monitoring node values over time that represent a current operation of the industrial asset, wherein the industrial asset is a physical machine and the plurality of monitoring nodes are sensors connected to the industrial asset;
• calculating a feature space version of current operation of the industrial asset;
• comparing, by the model adaptation element, the feature space version of augmented system model output with the feature space version of current operation of the industrial asset;
• and
• adapting parameters of the linear driven model based on a result of said comparison.