Automated creation of semantically-enriched diagnosis models

Patent metadata

Field	Value
Publication number	US-11734583-B2
Application number	US-201916688485-A
Country	US
Kind code	B2
Filing date	Nov 19, 2019
Priority date	Dec 30, 2013
Publication date	Aug 22, 2023
Grant date	Aug 22, 2023

Abstract

A system, method and a computer program product may be provided for automatically creating and parameterizing a semantically-enriched diagnosis model for an entity. The system receives a list of data points, from sensors or a database, to be used to create a diagnosis model. The system automatically creates the diagnosis model based on the received list of data points and data stored in a database and parameterizes the diagnosis model. The parameterized diagnosis model reflects rules that determine one or more potential causes of one or more abnormalities of one or more physical conditions in the entity.

First claim

The invention claimed is:

1. A method for automatically generating a parameterized semantic diagnosis model, the method comprising:
sensing, by using a network of sensors, a time series of data points for physical conditions associated with a specified system;
receiving, by a computer system, the time series of data points from the network of sensors;
semantically annotating, by the computer system, the received data points by adding semantic concepts to each of the data points, to create a semantic data point model comprised of a set of the semantically annotated data points;
accessing, by the computer system, one or more first-type rules from a knowledge base, the first-type rules describing which semantic concepts require a creation of physical variables;
creating, using said one or more first-type rules from the knowledgebase, variables representing the physical conditions detected by the sensors;
generating, by the computer system, a semantic variable model comprising a data representation of the created variables;
accessing, by the computer system, one or more second-type rules from the knowledge base, the second-type rules describing which variables correlate to or influence other physical variables;
determining, using the one or more second-type rules from the knowledgebase, a relationship between the created variables that reflect interactions between the created variables;
creating, by the computer system, a semantic model representing the determined relationship between the created variables that reflect said interactions;
accessing, by the computer system, one or more third-type rules from the knowledge base, the third-type rules describing which variables correlate or influence each other physical variables;
based on the created semantic model, creating, using said one or more third-type rules from the knowledgebase, a semantic diagnosis model to diagnose potential causes of the physical conditions of the specified system; and
identifying, by the computer system, based on the time series of data points from the network of sensors, one or more abnormal state instances of a semantic concept using an applied statistical technique;
parameterizing the semantic diagnosis model to create, based on statistics characterizing the time series of data points resulting from the applied statistical technique, a parameterized semantic diagnosis model to identify one or more potential causes of specified abnormal conditions of the specified system detected by the network of sensors; and
using said parameterized semantic diagnosis model to identify one or more potential causes of specified abnormal conditions of the specified system detected by the network of sensors.

2. The method according to claim 1 , wherein the semantic concepts include functionality represented by the received data points.

3. The method according to claim 1 , wherein the semantic concepts include measurements represented by the received data points.

4. The method according to claim 1 , wherein the semantic concepts include location information about the received data points.

5. The method according to claim 1 , wherein the determining a relationship between the created variables includes reasoning correlation directions between the created variables.

6. The method according to claim 1 , wherein the parameterized semantic diagnosis model maps one or more potential causes to the specified abnormal conditions.

7. The method according to claim 1 , wherein the parameterizing the semantic diagnosis model to create a parameterized semantic diagnosis model includes creating the parameterized semantic diagnosis model by analyzing historical time series of data from the sensors using statistical techniques.

8. The method according to claim 7 , wherein the analyzing historical time series of data from the sensors using statistical techniques includes computing a mean and a standard deviation for the historical time series of data.

9. The method according to claim 1 , wherein:
the parameterized semantic diagnosis model comprises a graph having nodes and edges connecting the nodes;
the nodes of the graph represent the created variables; and
the edges of the graph represent the determined relationships between the crcarcdcreated variables.

10. The method according to claim 9 , wherein:
each of the nodes of the graph has a value; and
the edges of the graph include correlation coefficients between the values of two of the nodes.

11. An apparatus for automatically generating a semantic diagnosis model, the apparatus comprising:
a memory device;
a processor coupled to the memory device, wherein the processor is configured to perform:
receive from a network of sensors a time series of data points for physical conditions associated with a specified system;
semantically annotate the received data points by adding semantic concepts to each of the data points, to create a semantic data point model comprised of a set of the semantically annotated data points;
access one or more first-type rules from a knowledge base, the first-type rules describing which semantic concepts require a creation of physical variables;
create, using said one or more first-type rules from the knowledgebase, variables representing the physical conditions detected by the sensors;
generate a semantic variable model comprising a data representation of the created variables;
access one or more second-type rules from the knowledge base, the second-type rules describing which variables correlate to or influence other physical variables;
determine, using the one or more second-type rules from the knowledgebase, a relationship between the created variables that reflect interactions between the created variables;
create a semantic model representing the determined relationship between the created variables that reflect said interactions;
access one or more third-type rules from the knowledge base, the third-type rules describing which variables correlate or influence each other physical variables;
based on the created semantic model, create, using said one or more third-type rules from the knowledgebase, a semantic diagnosis model to diagnose potential causes of the physical conditions of the specified system; and
identify, based on the time series of data points from the network of sensors, one or more abnormal state instances of a semantic concept using an applied statistical technique;
parameterize the semantic diagnosis model to create, based on statistics characterizing the time series of data points resulting from the applied statistical technique, a parameterized semantic diagnosis model to identify one or more potential causes of specified abnormal conditions of the specified system detected by the network of sensors; and
use said parameterized semantic diagnosis model to identify one or more potential causes of specified abnormal conditions of the specified system detected by the network of sensors.

12. The apparatus according to claim 11 , wherein the semantic concepts include functionality represented by the received data points.

13. The apparatus according to claim 11 , wherein the semantic concepts include location information about the received data points.

14. The apparatus according to claim 11 , wherein the determining a relationship between the created variables includes reasoning correlation directions between the created variables.

15. The apparatus according to claim 11 , wherein the parameterizing the semantic diagnosis model to create a parameterized semantic diagnosis model includes creating the parameterized semantic diagnosis model by comp

Assignees

• IBM

Inventors

• Lecue Freddy
• Ploennigs Joern
• Schumann Anika

Classifications

• G06N5/025 Primary

  Extracting rules from data · CPC title

• G06F16/367

  Ontology · CPC title