FDC Control Chart Warning Lines for Non-Normal Process Data

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current Fault Detection Classification (FDC) control charts lack statistical significance in warning line calculations, fail to account for non-normal data distributions, and require different computational logics for varying parameter characteristics, leading to potential false alarms and inadequate anomaly detection.

Innovation Solution

A method for generating control lines in FDC control charts involves clustering data into subgroups, calculating standard deviations, and adjusting warning line limits based on statistical analysis and parameter characteristics, ensuring the limits are statistically significant and adaptable to non-normal distributions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If the warning line is set based on SPC (Statistical Process Control) without statistical analysis, then the control chart can be established quickly, but the warning line lacks statistical significance and may produce false alarms

Engineering Contradiction:

Improvespeed of establishing control chartVSAvoidstatistical significance of warning line

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent transforms the warning line from a fixed SPC-based value to a dynamically calculated statistical threshold. It computes the warning line using the formula: warning_line = mean + k × standard_deviation, where k is a statistically determined multiplier based on the desired confidence level. This parameter transformation ensures the warning line has proper statistical significance while maintaining computational efficiency.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent replaces the empirical SPC methodology with a statistically rigorous approach. Instead of relying on conventional SPC rules of thumb, it substitutes a mathematically grounded method that calculates control limits based on the actual distribution characteristics of the process data, thereby eliminating false alarms while preserving quick establishment capability.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Ease of manufacture

If the control chart assumes normal distribution of data, then standard statistical methods can be applied, but false alarms occur when data does not follow normal distribution

Engineering Contradiction:

Improvesimplicity of statistical method applicationVSAvoidaccuracy of anomaly detection

Core Design Contradiction:

Ease of manufactureVSReliability

Solution Approach 1:

The patent introduces dynamic adaptation to the data distribution characteristics. It calculates the actual mean and standard deviation from the process data and uses these dynamic parameters to set the warning line, rather than relying on static normal distribution assumptions. This dynamic approach allows the control chart to adapt to various distribution types while maintaining statistical rigor.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The patent applies local statistical analysis by computing the mean and standard deviation specifically for the process data at hand, rather than assuming a global normal distribution. This localized statistical approach tailors the warning line calculation to the actual data characteristics, improving reliability for non-normally distributed data while keeping the method simple to implement.

Inventive Principle:

Principle #3Local quality

3Ease of operation

If a single computational logic is used for all parameter characteristics, then the method is simple to implement, but it cannot adequately handle diverse parameter data representations

Engineering Contradiction:

Improvesimplicity of computational logicVSAvoidhandling of diverse parameter characteristics

Core Design Contradiction:

Ease of operationVSAdaptability or versatility

Solution Approach 1:

The patent creates a universal computational framework that handles diverse parameter characteristics through a single unified method. The approach calculates the mean and standard deviation for any parameter type and applies the same warning line formula regardless of the specific parameter characteristics. This universal method eliminates the need for multiple computational logics while maintaining adaptability to different data types through its flexibility in accepting various input parameters.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20250053161A1Manufacturing control method and non-transitory computer readable media

Publication Date: 2025.02.13 UNITED MICROELECTRONICS CORP

AI summary

A manufacturing control method is applied to a computer system comprising a processor, a storage device, and a display device. The manufacturing control method includes: dividing a plurality of outlier-filtered data into a plurality of data subgroups based on a group division reference value; calculating a plurality of standard deviations for each of these data subgroups; calculating a warning line upper limit and a warning line lower limit based on the group division reference value, a predetermined multiple, and the standard deviations; adjusting either the warning line upper limit or the warning line lower limit based on the predetermined multiple and the standard deviations; and when a sensing data exceeds the warning line upper limit or the warning line lower limit, the computing system triggers a warning signal.