Manufacturing Process Anomaly Detection via Cycle-Time Hierarchies
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Complex industrial manufacturing processes face challenges in efficiently detecting anomalies and malfunctions due to the complexity of data from IIoT devices, making it difficult to measure overall equipment effectiveness (OEE) and Key Performance Indicators (KPIs) for quality assurance.
Innovation Solution
A method involving hierarchical structuring of the manufacturing process using a hierarchical tree based on cycle times of process variables, clustering, and anomaly detection through comparison with historic data to identify deviations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If hierarchical structuring and clustering of process variables is implemented, then anomaly detection efficiency is improved, but device complexity increases
Solution Approach 1:
The patent segments the complex manufacturing process data into hierarchical clusters based on cycle times. Process variables are grouped into clusters (e.g., clustering 300+ variables into manageable groups), and these clusters are arranged in hierarchical trees that organize data from most to least frequent cycle times. This segmentation transforms the overwhelming complex dataset into structured, manageable segments that can be efficiently analyzed for anomalies.
2Measurement precision
If cycle time-based clustering is performed on process variables, then measurement precision for anomaly detection is improved, but loss of information increases due to data aggregation
Solution Approach 1:
The patent applies local quality by treating different clusters with different levels of analysis based on their cycle time characteristics. High-frequency clusters (critical processes) are analyzed with higher precision and attention, while lower-frequency clusters receive appropriate but reduced analysis intensity. This ensures that measurement precision is optimized for critical anomalies without unnecessarily processing all data at maximum detail, thus balancing precision with information retention.
Solution Approach 2:
The patent transforms the analysis from a flat, single-dimensional view of all process variables to a multi-dimensional hierarchical structure organized by cycle time frequency. This dimensional transformation allows the system to preserve important temporal patterns and relationships while organizing data in a way that maintains critical information. The hierarchical tree structure adds a new organizational dimension that preserves variable relationships without losing essential process details.
3Ease of operation
If hierarchical tree structure is created from process variable clusters, then ease of operation for anomaly identification is improved, but device complexity increases
Solution Approach 1:
The hierarchical tree segments the complex set of process variables into organized clusters grouped by cycle time frequency. Each node in the tree represents a cluster of related variables, breaking down the overwhelming complexity into manageable hierarchical segments. This segmentation allows operators to navigate from high-level cluster summaries down to specific variables only when anomalies are detected, making the system easier to operate despite the underlying complexity.
Solution Approach 2:
The system performs preliminary clustering and hierarchical organization of all process variables before anomaly detection begins. This preliminary structuring based on cycle time patterns prepares the data in advance, so that during operation, anomalies can be quickly identified by comparing current data against the pre-established hierarchical structure, rather than searching through unorganized complex data in real-time.
Data Source
Figure 1
Figure 2a~3b
Figure 4~5
AI summary
The invention relates to the field of manufacturing process, particularly in an industrial plant, and to a computer-implemented method for detecting an anomaly of at least one device (10) in a manufacturing process. The method comprises the steps of: obtaining a time-series of a plurality of historic process variables (20) within a predefined time span; determining, for each process variable of the plurality of the historic process variables (20), a cycle time of the process variable; clustering the plurality of process variables into a plurality of clusters, wherein the process variables of a cluster have the same cycle time; arranging the plurality of clusters to a hierarchical tree (40), the tree being based on the cycle time of the clusters; storing the hierarchical tree (40); obtaining a time-series of a plurality of current process variables, which correspond to the historic process variables (20); and detecting the anomaly of the at least one device (10), wherein the anomaly is defined by identifying a cycle time of a current process variable that is longer than the cycle time of a corresponding historic variable.