Machine learning (ML)-based auto-visualization of plant assets

A method of monitoring assets of a process plant comprising:

• training a machine learning (ML) model on historical data associated with the process plant, the ML model being trained to determine damage mechanisms (DMs) corresponding to the assets of the process plant,
• wherein the ML model is trained on historical data that includes labelled data with various process parameters and corresponding DMs identified for each asset of the assets of the process plant;
• receiving data regarding the assets of the process plant from a plurality of data sources associated with the plant, the data regarding the assets includes design data of the assets and process parameter values of the assets;
• transforming the data regarding the assets into a standardized format;
• automatically identifying one or more corrosion loops (CLs) among the assets, wherein each of the one or more CLs includes a section of one or more of the assets that share common attributes;
• adding one or more properties to the assets, the added properties including at least an identification of corresponding CLs of the assets;
• determining, using the trained ML model, the DMs corresponding to each of the assets, wherein the DMs include causes of degradation of the assets;
• obtaining an equipment risk ranking for each of the assets based at least on the CLs, the DMs and the data associated with each of the assets;
• generating a recommended maintenance action for one or more of the assets based on corresponding equipment risk rankings for the assets;
• and
• generating one or more multi-dimensional visualizations of the assets where the multi-dimensional visualizations include overlays indicating the one or more CLs identified among the assets and the multi-dimensional visualizations are color coded based on the equipment risk rankings for each of the assets.

A non-transitory computer-readable storage medium comprising machine-readable instructions that cause a processor to:

• train a machine learning (ML) model on historical data associated with a process plant, the ML model being trained in determining damage mechanisms (DMs) that include causes of degradation of assets of the process plant,
• wherein the ML model is trained on historical data that includes labelled data with various process parameters and corresponding DMs identified for each asset of the assets of the process plant;
• receive data regarding the assets of the process plant from a plurality of data sources where the plurality of data sources include at least two-dimensional diagrams of the assets, the assets including equipment and lines of the process plant;
• automatically identify one or more corrosion loops (CLs) based at least on design data of the assets wherein each of the one or more CLs includes one or more of the assets with similar attributes and exposed to similar process conditions;
• retrieve process parameters associated with each of the assets within the process plant;
• automatically identify, by the ML model, at least one of the DMs associated with each of the assets based on the CLs and the process parameters associated with the CLs, wherein the DMs include causes of degradation of the assets;
• obtain an equipment risk ranking corresponding to each of the assets based on corresponding CLs, the DMs and current process conditions of each asset where the equipment risk ranking is representative of a probability of failure of the asset;
• and
• enable generation one or more multi-dimensional visualizations of the assets, wherein the multi-dimensional visualizations include overlays indicating the CLs identified among the one or more assets and the multi-dimensional visualizations display the assets with a color coding representative of the equipment risk ranking of the asset.