Electromechanical Fault Data Labeling Using Adaptive Time Windows
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data labeling methods for electromechanical devices are inaccurate due to delayed fault reporting, leading to mislabeling and ineffective training of prediction models, which are time-consuming and not scalable.
Innovation Solution
A method and apparatus for labeling data of electromechanical devices that estimate a second timestamp prior to a fault reporting event based on domain knowledge, iteratively adjust a time window, and determine an optimal evaluation metric to accurately label data as unhealthy or healthy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If data is labeled based on fault reporting events with timestamps, then the labeling process is automated and scalable, but the labeling accuracy deteriorates due to delayed fault reporting causing mislabeling
Solution Approach 1:
The system performs preliminary action by estimating the actual fault occurrence time before the fault reporting event is recorded. It calculates a time window extending backward from the reported fault time based on domain knowledge about typical fault detection delays, then labels data within this window as faulty. This preliminary labeling approach compensates for the delayed reporting and improves accuracy while maintaining automated scalability.
2Quantity of substance
If simulations and tests are performed to generate faulty data for labeling, then more faulty data is available for training, but the process becomes time-consuming, expensive, and less accurate in representing field data
Solution Approach 1:
The system applies self-service by automatically generating labeled faulty data from existing field operational data without requiring external simulations or manual tests. The automated time-window-based labeling method uses domain knowledge to identify and label faulty periods directly from operational data, eliminating the need for time-consuming simulation processes while maintaining representation of actual field conditions.
3Measurement precision
If the time window for labeling is increased to capture more faulty data, then more accurate labeling is achieved, but more healthy data is incorrectly labeled as unhealthy
Solution Approach 1:
The system applies partial action by using domain knowledge to determine an appropriate time window size that captures the majority of faulty data without excessively expanding the window. Rather than using a fixed large window that would capture all potential faulty data, the method uses a calculated window based on typical fault detection delays for the specific equipment type, achieving good accuracy while limiting false positives.
Solution Approach 2:
The system changes parameters by adjusting the time window size based on domain knowledge about specific equipment types and their fault detection characteristics. Different equipment types have different typical delays between fault occurrence and detection, so the labeling method adapts the time window parameter accordingly, optimizing the balance between capturing faulty data and avoiding false positives for each equipment category.
Data Source
AI summary
A method and an apparatus for labelling data of electromechanical devices include obtaining fault reporting event present in data associated with an electromechanical device; estimating second timestamp prior to first timestamp based on domain knowledge of electromechanical device; iteratively increasing a size of a time window by updating second timestamp; determining an evaluation metric of classifier in classifying data as unhealthy and healthy for each iteration; determining value of a second timestamp associated with optimal evaluation metric; and labelling data in time window formed between first timestamp and updated second timestamp as unhealthy data.