Sensor Derivative Drift Detection with Baseline Differential Distributions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Large and complex systems, such as infrastructure and manufacturing apparatuses, face challenges in monitoring numerous sensors effectively due to the difficulty in detecting anomalies and drifts in sensor measurement values, leading to potential economic losses and inefficiencies.
Innovation Solution
A data processing apparatus that calculates and compares first and second differential value sets from measurement data across different time periods, generating distributions and trend graphs to detect drifts in time derivatives, thereby improving monitoring efficiency and reducing erroneous detections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Difficulty of detecting and measuring
If machine learning models are used to detect anomalies by comparing predicted values with actual measurement values, then anomaly detection capability is improved, but the reliability of detection deteriorates due to unknown model deterioration and difficult update timing determination
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing differential value distributions during normal operation periods before anomalies occur. These pre-established baseline distributions are then used for comparison during monitoring, enabling reliable anomaly detection without requiring real-time model updates or retraining.
Solution Approach 2:
The patent introduces differential value distributions as an intermediary between raw sensor measurements and anomaly detection. By transforming measurement values into differential values and comparing them against reference distributions, the system mediates the detection process to achieve more reliable and interpretable anomaly detection.
2Measurement precision
If sensor measurement values are monitored for drift detection, then measurement accuracy is improved, but erroneous detection increases because time derivative information is not captured
Solution Approach 1:
The patent applies parameter changes by transforming the monitoring parameter from raw measurement values to differential values (time derivatives). This parameter transformation enables the system to detect drifts more accurately while avoiding erroneous detections, as the differential value distributions capture the rate of change information that is lost in instantaneous measurements.
3Reliability
If the number of sensors is increased to monitor all system components, then monitoring coverage is improved, but system complexity and monitoring difficulty increase
Solution Approach 1:
The patent extracts the essential monitoring information by focusing on differential value distributions rather than raw sensor data. This extraction approach allows the system to maintain comprehensive monitoring coverage while reducing the complexity of data processing and analysis, as the differential value distributions condense the information from multiple sensors into comparable statistical profiles.
Data Source
AI summary
According to one embodiment, a data processing apparatus includes a processor. The processor calculates, from the first measurement data, a first differential value set that is a set of first differential values in a time direction at a time included in the first period of the measurement values of the sensor of interest. The processor calculates, from the second measurement data, a second differential value set that is a set of second differential values in a time direction at a time included in the second period of the measurement values of the sensor of interest. The processor generates a first differential value distribution and a second differential value distribution using the second differential value set.