US 12,411,477 B2
Retrieving industrial asset data from disparate data sources
Prerna Juhlin, Heidelberg (DE); Somayeh Malakuti, Dossenheim (DE); Jens Doppelhamer, Ladenburg (DE); and Johannes Schmitt, Ladenburg (DE)
Assigned to ABB Schweiz AG, Baden (CH)
Filed by ABB Schweiz AG, Baden (CH)
Filed on Jul. 7, 2022, as Appl. No. 17/859,252.
Claims priority of application No. 21185901 (EP), filed on Jul. 15, 2021.
Prior Publication US 2023/0022369 A1, Jan. 26, 2023
Int. Cl. G05B 19/418 (2006.01)
CPC G05B 19/41865 (2013.01) [G05B 19/4183 (2013.01); G05B 19/4188 (2013.01)] 8 Claims
OG exemplary drawing
 
1. A computer-implemented method for retrieving data that relates to at least one industrial asset, comprising the steps of:
retrieving, from at least one data source relating to the at least one industrial asset, a set of types of data objects that the at least one data source provides;
for each type, determining, based on a synonym library and/or on contextual information of the industrial asset, at least one term that is to serve, besides this type, as a handle for accessing the respective data objects;
establishing a semantic association between the determined at least one term and the type from the set of types of data objects;
establishing a graph that represents the industrial asset, and/or an industrial plant that comprises the industrial asset, wherein vertices of the graph correspond to the set of types of data objects, and edges of the graph correspond to relationships between the data objects and the set of types of data objects:
generating, by a semantic enrichment component, semantic associations by linking the set of types of data objects to the at least one term by implementing, by the semantic enrichment component, an unsupervised learning technique on example instances of entity types to fill missing semantics and relationships with other entity types by invoking application programming interface (API) calls for object value retrieval;
clustering the set of types of data objects by means of the unsupervised learning technique:
in response to determining that a part of the types belonging to at least one cluster has a semantic association with at least one term, establishing a semantic association also between the remaining types of the set of types of data objects belonging to this cluster and at least one term, wherein a graph generator fills in missing edges of the graph using the unsupervised learning technique and the clusters of the set of types of data objects;
receiving, from a requesting entity, a query for the data objects from one or more lifecycle phases of the at least one industrial asset;
mapping, by a first software logic, the query to the set of types of data objects that are available from one or more given data sources relating to the at least one industrial asset, wherein each type of the set of types of data objects corresponds to a different value, range, spectrum, or structure;
obtaining, by a second software logic, from the one or more data sources, one or more data objects of the set of types of data objects;
producing, by the second software logic, from the one or more data objects, a response to the query; and transmitting the response to the requesting entity.