Sensor Change-Point Indexing for Quantifying Abnormality

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for analyzing time-series data from manufacturing sites struggle to quantify the degree of abnormality at change points, making it difficult to perform root cause analysis effectively.

Innovation Solution

A method and device that acquire sensor detection values, detect change points using a Sobel filter, and calculate an index value representing the degree of abnormality by comparing a target change point with a reference change point, allowing for quantitative assessment of abnormality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If existing analysis methods (self-regression model, neighbor method, specific spectral conversion method) are used to detect abnormality in time-series data, then abnormality detection capability is provided, but the degree of abnormality cannot be calculated quantitatively

Engineering Contradiction:

Improvequantitative measurement of abnormality degreeVSAvoidloss of quantitative abnormality information

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The invention changes the parameter representation by introducing an index value that quantifies the degree of abnormality. Instead of merely detecting whether abnormality exists, the system calculates a numerical index that represents the magnitude of abnormality at each change point, enabling quantitative comparison and analysis.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The invention replaces the qualitative assessment mechanism with a quantitative calculation mechanism. By using mathematical operations (calculating differences between consecutive data points, computing index values), the system substitutes subjective or qualitative abnormality assessment with objective numerical computation.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Difficulty of detecting and measuring

If change points are detected in sensor data to identify abnormalities, then abnormality detection capability is achieved, but effective root cause analysis becomes difficult without quantitative abnormality degrees

Engineering Contradiction:

Improvedifficulty of root cause analysisVSAvoidloss of abnormality severity information

Core Design Contradiction:

Difficulty of detecting and measuringVSLoss of information

Solution Approach 1:

The invention introduces feedback by calculating and providing the index value that quantifies abnormality degree. This quantitative feedback enables operators to understand not just that an abnormality occurred, but how severe it was, facilitating prioritization and root cause analysis by highlighting the most significant change points.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The invention transforms the parameter set by adding the index value parameter that represents abnormality degree. This additional parameter provides the information needed for effective root cause analysis, allowing differentiation between minor fluctuations and significant abnormalities that require immediate attention.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS11899439B2Method and device for calculating degree of abnormality

Publication Date: 2024.02.13 SEIKO EPSON CORP

AI summary

A method includes acquiring data indicating a plurality of sensor detection values arranged in order along a specific variable axis, detecting, in the data, a plurality of change points that are data points at which the sensor detection value on the specific variable axis changes by a predetermined value or more, and calculating, for a target change point that is one change point among the plurality of change points, an index value representing a degree of abnormality using the target change point and a reference change point that is another change point.