Control Cycle Abnormality Detection Using Internal State Features
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing predictive maintenance systems for equipment face delays in detecting abnormalities due to long data collection cycles and processing delays in control apparatuses, which hinder timely intervention.
Innovation Solution
A control apparatus that ensures consistent processing from state value collection to abnormality detection by using a feature quantity extraction unit to calculate feature quantities and an abnormality detection engine that evaluates the possibility of abnormalities based on learning data, with the ability to set arbitrary internal state values and determine unit sections for processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data aggregation is performed using a host data logger system, then data integration is achieved, but detection delay increases
Solution Approach 1:
The patent extracts the abnormality detection function from the centralized host data logger system and implements it directly in the control apparatus. This extraction eliminates the time-consuming data aggregation process at the host level, allowing real-time detection while maintaining reliable abnormality identification through local processing of control data.
Solution Approach 2:
The control apparatus performs preliminary abnormality detection processing directly on control data before the data would otherwise be aggregated at the host level. By conducting detection actions in advance at the control apparatus, the system eliminates subsequent aggregation delays and achieves immediate abnormality identification.
2Speed
If abnormality detection is implemented in a control apparatus, then detection speed improves, but processing consistency becomes difficult to ensure
Solution Approach 1:
The patent implements a feedback mechanism where the control apparatus continuously monitors control data and compares it against learned normal operation patterns. This feedback loop ensures consistent processing by constantly evaluating current state against established baselines, maintaining processing stability while enabling rapid detection of deviations.
Solution Approach 2:
The system utilizes parameter changes in control data over time to detect abnormalities. By monitoring how parameters evolve and comparing them against learned normal patterns, the system maintains processing consistency through systematic parameter analysis while achieving fast detection when parameters deviate from expected ranges.
3Loss of information
If multiple pieces of time series data are aggregated and integrated, then comprehensive analysis is achieved, but processing time increases
Solution Approach 1:
The patent extracts only the essential control data required for abnormality detection from the broader time series data, rather than aggregating and processing all available data. This selective extraction maintains sufficient information for accurate detection while dramatically reducing processing time by focusing only on relevant control parameters.
Solution Approach 2:
The system performs partial processing by focusing only on the specific control data elements necessary for abnormality detection, rather than comprehensively processing all time series data. This partial action approach achieves adequate detection capability with minimal processing time by avoiding unnecessary data aggregation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided is a mechanism for integrated processing of all stages from collection of state values from an object being controlled to error detection processing in a control device. The control device comprises: a management unit that updates internal state values for each control cycle, generated in the object being controlled; a feature quantity extraction unit that calculates the feature quantity from change in the second internal state value for each unit sector determined according to the value for the first internal state value; and an abnormality detection unit that generates detection results indicating whether an abnormality has occurred in the monitoring subject contained in the object being controlled, on the basis of the feature quantity. The feature quantity extraction unit and the abnormality detection unit output the values for the first internal state value used to determine the unit section, in correspondence with the corresponding unit section.