Back to results
US 2026/0288131 A1Published application
Interactive alarm
AssigneeYokogawa Electric Corporation
InventorsAiswarya PRASANNAKUMAR, Bijuraj PANDIYATH VELAYUDHAN
Life of this patent
Estimating the term…
Abstract
Industrial sites utilize process components, or "tags," to perform, monitor, and control industrial processes. When an alarm occurs, it can be critical to quickly execute an action to best resolve the alarm. Providing an operator with a prioritized list of recommended actions to a particular alarm enables the user to quickly identify and take the recommended action to resolve the alarm. The recommendation may be determined by an artificial intelligence, such as a trained neural network, analyzing process components and their interrelationships associated with an alarm.
Claims20
A computer-implemented method of training a neural network for alarm resolution actions, comprising:
- collecting a set of past operational attributes of an industrial process;
- applying one or more transformations to each of the past operational attributes of the set of past operational attributes of the industrial process including altering the degree of a first transformation of the one or more transformations, altering the order of the one or more transformations applied to two or more past operational attributes, including a second transformation to an omitted one of the past operational attributes of the set of past operational attributes, and omitting at least one transformation of the one or more transformations to an included one of the past operational attributes to create a modified set of past operational attributes;
- creating a first training set comprising the collected set of past operational attributes, the modified set of past operational attributes, and a set of extraneous operational attributes of the industrial process not associated with a prior alarm;
- training the neural network in a first stage using the first training set;
- creating a second training set for a second stage of training comprising the first training set and the set of extraneous operational attributes of the industrial process not associated with any prior alarms incorrectly identified as impacting the prior alarm;
- and
- training the neural network in the second stage using the second training set.
4 more elements
A computer-implemented method, comprising:
- collecting a set of operational attributes of an industrial process;
- analyzing the set of operational attributes to identify subsets of operational attributes being interdependent;
- receiving an alarm for the industrial process;
- and
- in response to the alarm:
- identifying a subset of operational attributes associated with the alarm;
- identifying a first operational attribute, of the subset of operational attributes associated with the alarm, triggering the alarm;
- identifying a second operational attribute interdependent on the first operational attribute;
- presenting indicium of the second operational attribute on a graphical user interface (GUI) and receiving a modification input therefrom;
- and
- modifying the second operational attribute in accordance with the modification input.
8 more elements
A system, comprising:
- a database;
- and
- a processor, coupled to a computer memory maintaining thereon instructions to cause the processor to execute the instructions;
- and
- wherein the instructions cause the processor to perform:
- collecting a set of past operational attributes of an industrial process from the database;
- applying one or more transformations to each of the past operational attributes of the set of past operational attributes including altering the degree of a first transformation of the one or more transformations, altering the order of the one or more transformations applied to two or more past operational attributes, including a second transformation to an omitted one of the past operational attributes of set of past operational attributes, and omitting at least one transformation of the one or more transformations to an included one of the past operational attributes of the set of past operational attributes to create a modified set of past operational attributes;
- creating a first training set comprising the collected set of past operational attributes, the modified set of past operational attributes, and a set of extraneous operational attributes of the industrial process not associated with a prior alarm;
- training a neural network in a first stage using the first training set;
- creating a second training set for a second stage of training comprising the first training set and the set of extraneous operational attributes of the industrial process not associated with any prior alarms incorrectly identified as impacting the prior alarm;
- and
- training the neural network in the second stage using the second training set.
9 more elements
Dependencies are inferred from the claim text.