Sensor Time-Series Synchronization for Manufacturing Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Time series of sensor values from manufacturing processes often become shifted relative to each other, making it difficult to synchronize them for effective anomaly detection.
Innovation Solution
A method that detects deviations in sensor time series from corresponding points in a reference time series, shifts the time series accordingly, and uses dynamic time warping to synchronize them, without requiring hyperparameter definition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If time series of sensor values are detected during manufacturing process executions, then anomaly detection capability is improved, but time series synchronization becomes problematic due to temporal shifts
Solution Approach 1:
The patent applies preliminary action by performing synchronization of time series data before anomaly detection. The system identifies characteristic points in each time series, calculates temporal deviations from reference points, and shifts time series accordingly to align them in advance, ensuring proper temporal alignment before the actual anomaly detection process begins
Solution Approach 2:
The patent uses characteristic points as intermediaries to achieve synchronization. These characteristic points serve as reference markers that mediate between different time series, allowing the system to calculate deviations and apply appropriate shifts to align multiple time series without directly comparing every data point
2Loss of time
If traditional synchronization methods are used, then temporal alignment is achieved, but system complexity increases due to hyperparameter definition requirements
Solution Approach 1:
The patent applies self-service by enabling the system to automatically identify characteristic points and calculate synchronization shifts without requiring external configuration or hyperparameter definition. The system autonomously determines the temporal relationships between time series based on the inherent structure of the data, eliminating the need for manual parameter tuning
Solution Approach 2:
The patent uses a reference time series as a template or copy that represents the ideal temporal structure. Other time series are synchronized by comparing their characteristic points against this reference, copying the temporal pattern from the reference series to align all series without complex configuration
3Loss of time
If time series are shifted for synchronization, then temporal alignment is improved, but original time series characteristics may be altered
Solution Approach 1:
The synchronization is performed as a preliminary step before anomaly detection, creating aligned time series that preserve the original characteristics within each series while achieving proper temporal relationships between series. The characteristic points and their relative positions within each time series remain intact
Solution Approach 2:
Characteristic points serve as intermediaries that maintain the structural integrity of each time series during synchronization. By aligning based on these key points rather than shifting every data point uniformly, the method preserves the local characteristics and patterns within each time series while achieving global temporal alignment
Data Source
AI summary
A method for synchronizing time series of sensor values relating to a manufacturing process. The method includes: detecting at least two time series of sensor values by the same sensor during each of at least two executions of the manufacturing process; ascertaining, for each of the at least two time series of sensor values, deviations of different points of the corresponding time series from corresponding points of a reference time series and shifting the corresponding time series by a value ascertained based on the ascertained deviations, in order to synchronize the time series; and providing the synchronized time series.