Monitoring and controlling an operation of a distillation column

A device, comprising:

• one or more memories;
• and
• one or more processors, communicatively coupled to the one or more memories, configured to:
• process, using a clustering technique, historical data to identify a plurality of operating states of a distillation column;
• determine, based on a first set of parameters from the historical data, occurrences of a hazardous condition during a historical time period while the distillation column was operating according to an operating state of the plurality of operating states;
• generate a prediction model for monitoring the distillation column for the hazardous condition;
• train the prediction model based on training data that is associated with the occurrences of the hazardous condition,
• wherein the training data includes historical sensor data and historical control data that is associated with a second set of parameters;
• monitor, using the prediction model, the distillation column to determine a probability that the distillation column is going to experience the hazardous condition within a threshold time period,
• wherein the prediction model is configured to determine the probability based on measurements from a set of sensors of the distillation column;
• and
• perform, based on the probability satisfying a probability threshold, an action associated with the distillation column.

A system comprising:

• a distillation column;
• a plurality of sensors to monitor a flow of fluid through the distillation column;
• one or more control devices to control the flow of fluid through the distillation column;
• and
• a controller that is configured to:
• obtain historical data associated with the distillation column;
• process the historical data to identify a plurality of operating states of the distillation column;
• determine, based on a set of parameters from the historical data, occurrences of a hazardous condition during a historical time period while the distillation column was operating according to an operating state of the plurality of operating states;
• generate a prediction model to monitor the distillation column for the hazardous condition;
• train the prediction model based on training data that is associated with the occurrences of the hazardous condition;
• and
• perform an action associated with monitoring the distillation column using the prediction model.