Power Converter Warning Analysis for Early Fault Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current monitoring systems for power converters and drives generate a large number of warning messages, making it difficult to predict and prevent potential faults, especially in high-power applications where downtime can be costly and disruptive.
Innovation Solution
A method that uses a combination of user-defined and system-dependent warning messages, grouped and analyzed using algorithms to predict impending errors, with the aid of artificial intelligence, allowing for timely countermeasures and reducing the number of messages to focus on critical issues, thereby improving monitoring accuracy and reducing downtime.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a large number of warning messages are generated to monitor all potential issues, then monitoring coverage is improved, but the ability to predict and prevent faults deteriorates due to information overload
Solution Approach 1:
The monitoring system segments warning messages into different categories (critical warnings, non-critical warnings, informational messages) and applies different analysis methods to each segment. Critical warnings are analyzed in real-time using AI algorithms, while non-critical messages are aggregated and analyzed periodically, allowing the system to maintain comprehensive monitoring coverage while preventing information overload in the prediction process
Solution Approach 2:
An AI-based intermediary analysis layer is introduced between the raw warning messages and the fault prediction process. This intermediary automatically filters, prioritizes, and correlates warnings using machine learning models, extracting meaningful patterns from the large volume of messages while eliminating noise, thus preserving the ability to predict faults despite comprehensive monitoring
2Measurement precision
If all warning messages are analyzed individually to ensure accurate fault prediction, then prediction accuracy is improved, but processing time and system complexity worsen
Solution Approach 1:
The system merges multiple warning messages that indicate similar or related issues into single consolidated analysis units. For example, multiple temperature-related warnings from different components are combined into a unified thermal analysis, maintaining prediction accuracy by preserving the collective information while reducing the total number of individual analyses required
Solution Approach 2:
The AI analysis system dynamically changes analysis parameters based on the type and severity of warnings. For critical warnings, the system applies detailed analysis with high computational resources, while for minor warnings, it uses simplified analysis models, thus maintaining accuracy for critical predictions while reducing overall system complexity and processing requirements
3Loss of time
If comprehensive warning analysis is performed to predict faults early, then downtime reduction is improved, but the number of false alarms increases
Solution Approach 1:
The system implements feedback loops where AI algorithms continuously learn from the outcomes of previous predictions. When a predicted fault does not materialize (false alarm), the system adjusts its prediction thresholds and parameters. When predictions are confirmed accurate, the models are reinforced, thereby reducing false alarms over time while maintaining early detection capabilities that minimize downtime
Solution Approach 2:
The system performs preliminary correlation and validation of warning patterns before generating fault predictions. Multiple warnings must meet specific temporal and logical criteria before triggering a prediction, which filters out spurious signals while maintaining sensitivity to genuine fault precursors, thus reducing false alarms without compromising downtime reduction
4Adaptability or versatility
If user-defined warning messages are allowed to customize monitoring, then adaptability is improved, but system complexity and configuration difficulty worsen
Solution Approach 1:
The system provides a universal set of pre-configured warning templates and analysis patterns that can be applied across different power converter applications. Users can select and activate relevant templates based on their specific needs without having to create custom configurations from scratch, thereby maintaining high adaptability while simplifying the configuration process through reusable multi-functional building blocks
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
In a method for monitoring a converter (1), a plurality of warning messages is used to conclude whether there is a failure in the converter (1). For this purpose, warning messages (2) are grouped according to the parts of the converter (1) affected.