Fault Elimination Control Using Root Cause and Virtual Mitigation Testing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current machine learning model-based systems for fault detection in technical installations lack the ability to accurately identify the root cause of faults and do not test the impact of mitigation actions before implementation, leading to potential incorrect shutdowns and inefficiencies.

Innovation Solution

A method and system that determine the root cause of faults and predict the outcome of mitigation actions using sensor data, machine learning models, and virtual models of the technical installation, allowing for informed decision-making before implementation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If machine learning model-based systems are used for fault detection, then continuous evaluation of parameters and alarm generation is achieved, but the system may incorrectly identify faults due to insufficient training data, leading to shutdown of the technical installation

Engineering Contradiction:
Improvecontinuous evaluation capabilityVSAvoidfault identification accuracy
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The system performs preliminary actions by generating multiple candidate mitigation actions before implementing any single action. The virtual model tests these candidate actions in advance to predict their outcomes, ensuring that the selected mitigation action is validated before actual implementation in the technical installation.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

A virtual model (digital twin) of the technical installation is created as a copy to test mitigation actions. This virtual copy allows the system to evaluate potential mitigation actions and predict their outcomes without affecting the actual technical installation, thereby improving reliability of fault resolution.

Inventive Principle:
Principle #26Copying

2Loss of time

If machine learning model-based systems provide mitigation actions without testing, then quick response to faults is achieved, but there is no way of predicting whether the mitigation action would generate a preferred outcome

Engineering Contradiction:
Improveresponse time to faultVSAvoidoutcome prediction capability
Core Design Contradiction:
Loss of timeVSLoss of information

Solution Approach 1:

The system performs preliminary actions by generating multiple candidate mitigation actions before implementing any single action. The virtual model tests these candidate actions in advance to predict their outcomes, ensuring that the selected mitigation action is validated before actual implementation in the technical installation.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The virtual model provides feedback by simulating the technical installation's response to candidate mitigation actions. This feedback mechanism allows the system to evaluate whether a mitigation action would produce the desired outcome before implementing it in the actual installation, thereby reducing uncertainty.

Inventive Principle:
Principle #23Feedback

3Productivity

If the system implements mitigation actions without virtual model testing, then immediate fault resolution is achieved, but time and resources may be wasted on ineffective actions

Engineering Contradiction:
Improvefault resolution speedVSAvoidresource wastage
Core Design Contradiction:
ProductivityVSLoss of energy

Solution Approach 1:

The system performs preliminary actions by generating multiple candidate mitigation actions before implementing any single action. The virtual model tests these candidate actions in advance to predict their outcomes, ensuring that the selected mitigation action is validated before actual implementation in the technical installation.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3660612B1Method and system for elimination of fault conditions in a technical installation
Publication Date: 2022.12.28 SIEMENS AG
  • EP3660612B1 patent drawingFigure 1
  • EP3660612B1 patent drawingFigure 2
  • EP3660612B1 patent drawingFigure 3~4

AI summary

A method (300) and system (100) for eliminating a fault condition in a technical installation (107) is provided. The method includes predicting an occurrence of the fault condition in at least a portion of the technical installation, determining a root cause of the predicted fault condition and identifying one or more mitigation actions to resolve the fault condition. Furthermore, the method includes determining an outcome associated with at least one of the one or more mitigation actions on the technical installation (107) and outputting on a device (108) associated with a user at least one mitigation action to be implemented in the technical installation (107) based on the determined outcome.