Systems and methods for real-time root cause analysis in industrial processes

Patent metadata

Field	Value
Publication number	US-2024142962-A1
Application number	US-202017754923-A
Country	US
Kind code	A1
Filing date	Oct 19, 2020
Priority date	Oct 18, 2019
Publication date	May 2, 2024
Grant date	—

Abstract

Conventionally, root cause analysis and process documentation in process industries has been manually performed resulting in time consuming effort, cost, and human resources. Moreover, in the event of failure, looking at such document and searching for possible root causes is practically impossible in the interest of time and cost associated. Systems and methods of the present disclosure systematically curate knowledge of industrial process(es) from various sources and generate process ontology via meta model(s). Root cause graph (RCG) is created wherein the RCG corresponds to process and root cause and failure modes in the process. The RCG is then transformed to machine instructions which are executed for root cause analysis in real time. The created graphs/knowledge also help in identifying conflicting knowledge or redundant knowledge. Present disclosure enables root cause analysis as soon as a failure occurs or as the systems show or indicate a tendency towards failure.

First claim

1 . A processor implemented method for real-time root cause analysis, comprising:
obtaining, via one or more hardware processors, information pertaining to one or more industrial processes being executed by one or more equipment, wherein the information comprises one or more of piping and instrumentation diagram (PID), operational data, maintenance history, root cause knowledge, or a corresponding process model; generating, via one or more meta models executed by the one or more hardware processors, a process ontology using the obtained information; transforming the root cause knowledge to a set of machine instructions, the set of machine instructions comprising an associated detection state, wherein a logic in the set of machine instruction is derived from the root cause knowledge, wherein the one or more detections are identified through a detection model using data captured in real time by the one or more sensors, and wherein information from the one or more sensors is mapped to the process ontology; and generating, via the one or more hardware processor, a root cause path comprising one or more root causes and associated interdependencies using (i) the process ontology and (ii) the root cause knowledge, wherein the one or more root causes are detected using the detection model, and wherein the root cause path further comprises one or more actions to be executed.

2 . The processor implemented method of claim 1 , wherein the root cause knowledge is based on one or more of the PID, the operational data, the maintenance history, domain knowledge, and the corresponding process model.

3 . The processor implemented method of claim 1 , wherein the corresponding process model is at least one of a data-based model, a physics-based model, an empirical model, and a hybrid model.

4 . The processor implemented method of claim 1 , wherein at least one of a conflicting knowledge or a redundant knowledge is identified from the root cause graph.

5 . The processor implemented method of claim 1 , wherein the detection model is at least one of a data-based model, a physics-based model, a pattern identification model, an empirical model, and a hybrid model.

6 . The processor implemented method of claim 1 , further comprising generating, via the one or more hardware processor, a root cause graph using the process ontology and the root cause knowledge, wherein the root cause graph comprises one or more corresponding root causes detected using the detection model.

7 . The processor implemented method of claim 6 , wherein a first root cause comprised in the root cause graph is indicative of a performance indicator (PI) corresponding to the one or more industrial processes, wherein one or more root causes are hierarchical arranged after one or more performance indicators (PIs) in the root cause graph, and wherein the root cause graph is represented in at least one of a tree representation format, a tabular representation format, a graphical representation format, and an unstructured representation format.

8 . The processor implemented method of claim 1 , wherein the one or more corresponding root causes are detected based on a unique combination of one or more detections and the associated detection state thereof, using the detection model, and wherein the associated detection state is at least one of a positive detection state and a negative detection state.

9 . The processor implemented method of claim 1 , wherein the generated process ontology comprises information pertaining to one or more of (i) the one or more equipment, (ii) a corresponding location of one or more sensors deployed within the one or more equipment, (iii) sensory information captured through the one or more sensors thereof, (iv) details specific to an interaction between at least one of (a) the one or more equipment and (b) the one or more industrial processes, (v) one or more parameters of the one or more industrial processes, or (vi) one or more action plans, one or more detection states, and wherein the one or more action plans comprise one or more of repair, mitigation, containment, or control of at least one of the one or more industrial processes and the one or more equipment.

10 . A system, comprising:
a memory storing instructions; one or more communication interfaces; and one or more hardware processors coupled to the memory via the one or more communication interfaces, wherein the one or more hardware processors are configured by the instructions to: obtain information pertaining to one or more industrial processes being executed by one or more equipment, wherein the information comprises one or more of piping and instrumentation diagram (PID), operational data, maintenance history, root cause knowledge, or a corresponding process model; generate, via one or more meta models, a process ontology using the obtained information; transform the root cause knowledge to a set of machine instructions, the set of machine instructions comprising an associated detection state, wherein a logic in the set of machine instruction is derived from the root cause knowledge, wherein the one or more detections are identified through a detection model using data captured in real time by the one or more sensors, and wherein information from the one or more sensors is mapped to the process ontology; and generate a root cause path comprising one or more root causes and associated interdependencies using (i) the process ontology and (ii) the root cause knowledge, wherein the one or more root causes are detected using the detection model, and wherein the root cause path further comprises one or more actions to be executed.

11 . The system of claim 10 , wherein the root cause knowledge is based on one or more of the PID, the operational data, the maintenance history, domain knowledge, and the corresponding process model.

12 . The system of claim 10 , wherein the corresponding process model is at least one of a data-based model, a physics-based model, an empirical model, and a hybrid model

13 . The system of claim 10 , wherein at least one of a conflicting knowledge or a redundant knowledge is identified from the root cause graph.

14 . The system of claim 10 , wherein the detection model is at least one of a data-based model, a physics-based model, a pattern identification model, an empirical model, and a hybrid model.

15 . The system of claim 10 , wherein the one or more hardware processors are further configured by the instructions to generate a root cause graph using the process ontology and the root cause knowledge, wherein the root cause graph comprises one or more corresponding root causes detected using the detection model.

16 . The system of claim 15 , wherein a first root cause comprised in the root cause graph is indicative of a performance indicator (PI) corresponding to the one or more industrial processes, wherein one or more root causes are hierarchical arranged after one or more performance indicators (PIs) in the root cause graph, and wherein the root cause graph is represented in at least one of a tree representation format, a tabular representation format, and a graphical representation format.

17 . The system of claim 10 , wherein the one or more corresponding root causes are detected based on a unique combination of one or more detections and the associated detection state thereof, using the detection model, and wherein the associated detection state is at least one of a positive detection state and a negative detection state.

18 . The system of claim 10 , wherein the generated process ontol

Assignees

• Tata Consultancy Services Ltd

Inventors

• Ahmad Dilshad
• Basavarsu Purushottham Gautham
• Kulkarni Hrishikesh Nilkanth
• Malhotra Chetan Premkumar
• Kalidoss Thanga Jawahar
• Kolappan Swamy Doss

Classifications

• G05B23/0275 Primary

  Fault isolation and identification, e.g. classify fault; estimate cause or root of failure · CPC title

• G06Q10/20 Primary

  Administration of product repair or maintenance · CPC title

• G05B23/0281

  Quantitative, e.g. mathematical distance; Clustering; Neural networks; Statistical analysis · CPC title