Industrial Anomaly Diagnosis Using Time-Segmented Sensor Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional anomaly detection and diagnosis techniques in industrial processes fail to effectively account for the temporal behavior of sensor data, limiting their applicability in identifying anomalies in manufacturing and process industries.
Innovation Solution
A processor-implemented method and system that processes multivariate time series data by dividing it into segments, extracting features using an encoding mechanism, reconstructing data using decoding, and calculating reconstruction error to identify anomalous segments and faulty sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional anomaly detection techniques are used, then the detection process is simple, but the detection precision is insufficient due to ignoring temporal behavior
Solution Approach 1:
The patent segments the multivariate time series data into multiple segments along the temporal dimension, allowing the system to capture temporal behavior patterns while maintaining manageable computational complexity. Each segment is processed independently to identify anomalies, improving detection precision without overwhelming system resources.
Solution Approach 2:
The patent transforms the temporal analysis by dividing data along the time dimension into segments, effectively adding a segment dimension to the analysis. This dimensional transformation enables the system to capture temporal dependencies while maintaining computational efficiency through localized processing.
2Measurement precision
If individual data points are treated as independent, then the processing speed is fast, but the detection accuracy deteriorates due to ignoring temporal dependencies
Solution Approach 1:
By segmenting the time series data into fixed-size windows, the patent enables parallel processing of multiple segments simultaneously. This approach captures temporal dependencies within each segment while allowing efficient batch processing across segments, balancing accuracy and processing time.
Solution Approach 2:
The patent processes only relevant features within each segment rather than analyzing all data points in their entirety. This partial action approach focuses computational resources on extracting meaningful temporal patterns, achieving high detection accuracy with reduced processing time.
3Measurement precision
If feature extraction is performed at multiple stages, then the detection precision improves, but the computational complexity increases
Solution Approach 1:
The patent performs feature extraction at multiple stages within each segmented window rather than processing the entire dataset at once. This segmentation enables progressive feature extraction, where important features are identified at each stage, improving precision while managing computational energy through localized processing.
Solution Approach 2:
The patent performs preliminary feature extraction at multiple stages within segments before final anomaly detection. This preliminary action identifies and extracts important temporal patterns early, reducing the computational burden of subsequent analysis while maintaining high detection precision.
Data Source
AI summary
Industrial processes and equipment are prone to operational changes and faulty operation of such processes and equipment can adversely affect output of the overall setup. Existing systems for monitoring and fault detection consider individual instances of data for fault detection, which may not be suitable for industrial processes. Disclosed herein is a system and a method for anomaly detection in an industrial enterprise. The system collects data from a plurality of sensors as input. The system processes the collected data along temporal dimension, during which the data is split to multiple segments of fixed window size. Data in each segment is processed to identify anomalous data, and data in segments identified as containing the anomalous data is further processed to identify one or more sensors that are faulty and are contributing to the anomalous data.