Industrial Alarm Management Using ANN Criticality Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Complex industrial processes often present unclear behavior to operators and service personnel, making it difficult to recognize abnormal behavior, especially due to the overwhelming number of alarms generated, which can lead to confusion and errors.
Innovation Solution
A method utilizing a machine learning model, specifically an artificial neural network (ANN), is trained with input data including time-series of observable process-values, manipulated variables, and internal variables, to predict abnormal behavior and output criticality values, thereby aiding in alarm management and reducing the complexity of understanding industrial process criticality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional alarm management systems are used to monitor industrial processes, then comprehensive process monitoring is achieved, but the number of alarms becomes overwhelming and difficult to manage
Solution Approach 1:
The alarm management system segments alarms by criticality levels (e.g., high, medium, low) and categories, allowing operators to focus on the most critical issues first. This segmentation reduces the overwhelming nature of comprehensive alarm lists by organizing them into manageable groups with distinct priority levels.
Solution Approach 2:
The system introduces an intermediary intelligence layer that analyzes alarm data, process data, and contextual information to generate synthesized alarm summaries and recommendations. This intermediary processing layer filters and prioritizes alarms before presenting them to operators, reducing the direct burden of raw alarm volume.
2Reliability
If multiple alarms are raised to capture all critical events, then comprehensive safety coverage is achieved, but operator mental load increases and recognition of abnormal behavior becomes difficult
Solution Approach 1:
The system merges multiple related alarms into consolidated alarm summaries or single representative alarms when they indicate the same underlying issue. Related alarms are combined with their root cause identified, reducing the total number of separate alarm notifications while maintaining comprehensive safety coverage through the merged information.
Solution Approach 2:
The system provides feedback to operators by analyzing alarm patterns and providing contextual information about abnormal behaviors. This feedback includes recommendations, trend analysis, and causal relationships that help operators understand the significance of alarms without being overwhelmed by raw alarm counts.
3Speed
If traditional alarm systems provide immediate alerts for all process deviations, then rapid response to hazards is achieved, but the clarity of critical information is reduced due to alarm flooding
Solution Approach 1:
The system performs preliminary analysis and prioritization of alarms before they reach operators. By pre-processing alarm data, identifying critical patterns, and ranking alarms by significance, the system ensures that the most critical information is presented first, maintaining rapid response capability while preserving information clarity through pre-organized prioritization.
Data Source
AI summary
A method and computer program product including training a machine learning model by means of input data and score data, wherein the machine learning model is an artificial neural net, ANN; running the trained machine learning model by applying the first time-series to the trained machine learning model; and outputting, by the trained machine learning model, an output value, comprising at least a second criticality value of the at least one predicted observable process-value indicative of the abnormal behaviour of the industrial process in a predefined temporal distance.