Plant Operating Mode Identification for Abnormal Event Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current industrial plant operations lack data-driven and readily consumable information to determine and diagnose the cause of events, making it difficult to identify the state and operating mode, especially during maintenance, startup, and shutdown states, which are critical for maintaining stability and safety.
Innovation Solution
A self-configuring, self-updating machine learning-based artificial intelligence algorithm is implemented to determine the state and operating mode of industrial plants by detecting changes in manipulated parameters, utilizing multiple data sources such as sensor data, human machine interface graphics, operation logs, and engineering configuration data, without requiring human input, and providing real-time guidance to operators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional plant operation monitoring is used, then the system is simple to operate, but it cannot accurately determine plant state and operating mode during critical events
Solution Approach 1:
The machine learning model automatically determines plant state and operating mode by processing sensor data without requiring manual operator input or configuration. The system self-configures and self-updates by continuously learning from historical data, eliminating the need for complex manual setup while achieving accurate state determination during critical events.
Solution Approach 2:
The patent replaces traditional manual monitoring and mechanical decision-making processes with an artificial intelligence-based machine learning system. This substitution enables automated analysis of multiple sensor data sources to accurately identify plant state and operating mode, significantly improving measurement precision while managing complexity through software-based solutions.
2Measurement precision
If machine learning-based AI algorithm is implemented, then state determination accuracy is improved, but data processing complexity increases
Solution Approach 1:
The system performs preliminary actions by continuously training the machine learning model with historical sensor data before actual state determination is needed. This pre-processing and pre-training approach allows the model to be ready for accurate real-time state determination during critical events, managing complexity through advance preparation.
Solution Approach 2:
The machine learning model incorporates feedback mechanisms by continuously learning from new data and updating its parameters. This feedback loop improves state determination accuracy over time while the system automatically manages the complexity of data processing through iterative optimization rather than manual intervention.
3Reliability
If real-time monitoring of multiple data sources is performed, then event prediction capability is improved, but information processing load increases
Solution Approach 1:
The system extracts only the most relevant features and patterns from multiple sensor data sources that are critical for predicting events and determining plant state. By focusing on key indicators rather than processing all raw data equally, the system improves event prediction capability while reducing unnecessary computational energy consumption.
Data Source
AI summary
A system processes historical facility data that relate to facility states and modes of operation. The historical facility data are clustered into groups representing the facility states and the modes of operation. The groups are used to determine a current state and mode of the facility. When the facility is in a normal state, the system determines whether an event in the facility is an abnormality. If an abnormality is identified, the system transmits a signal indicating the abnormality.