MONITORING AND CONTROLLING AN OPERATION OF DISTILLATION COLUMN

In some implementations, a control system may obtain historical data associated with usage of a distillation column during a historical time period. The control system may configure a prediction model to monitor the distillation column for a hazardous condition. The prediction model may be trained based on training data that is associated with occurrences of the hazardous condition. The control system may monitor, using the prediction model, the distillation column to determine a probability that the distillation column experiences the hazardous condition within a threshold time period. The prediction model may be configured to determine the probability based on measurements from a set of sensors of the distillation column. The control system may perform, based on the probability satisfying a probability threshold, an action associated with the distillation column to reduce a likelihood that the distillation column experiences the hazardous condition within the threshold time period.