Model-Plant Mismatch Detection via Parameter Clustering

Solution Overview

Problem

Model-based industrial process controllers face challenges in determining whether their models accurately represent the true behavior of industrial processes, especially in changing conditions, due to the difficulty of using routine operating data and the undesirability of performing experiments that may affect product quality.

Innovation Solution

A method for detecting model-plant mismatch using model parameter data clustering, which involves repeatedly identifying model parameter values, clustering them, and identifying additional values to detect mismatches by determining if new values fall outside the clusters, allowing for updates or creation of new models to maintain product quality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If experiments are performed to determine model accuracy, then model-plant mismatch detection capability is improved, but product quality may be affected and production is disrupted

Engineering Contradiction:

Improvemodel accuracy determinationVSAvoidproduct quality

Core Design Contradiction:

Measurement precisionVSManufacturing precision

Solution Approach 1:

The patent applies preliminary action by pre-processing routine operating data to extract model parameter values and clustering them into reference clusters before actual mismatch detection. This preparation work is done using normal operational data, so when mismatch detection is needed, no additional experiments disrupting production are required - the detection can proceed immediately using the pre-established clusters, thus maintaining both detection capability and product quality.

Inventive Principle:

Principle #10Preliminary action

2Productivity

If routine operating data is used for model parameter identification, then production continuity is maintained, but the ability to detect significant model-plant mismatches is insufficient

Engineering Contradiction:

Improveproduction continuityVSAvoidmismatch detection capability

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent applies preliminary action by pre-processing routine operating data to extract model parameter values and clustering them into reference clusters before actual mismatch detection. This preparation work is done using normal operational data, so when mismatch detection is needed, no additional experiments disrupting production are required - the detection can proceed immediately using the pre-established clusters, thus maintaining both detection capability and product quality.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent applies feedback by comparing newly identified model parameter values against the pre-established reference clusters and using the mismatch detection results to trigger model updates. This closed-loop feedback mechanism ensures that routine operating data continuously contributes to improving mismatch detection capability without interrupting production, as the system learns and adapts from ongoing operational data.

Inventive Principle:

Principle #23Feedback

3Measurement precision

If model parameters are repeatedly identified and clustered, then mismatch detection accuracy is improved, but computational complexity increases

Engineering Contradiction:

Improvemismatch detection accuracyVSAvoidcomputational complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies preliminary action by pre-processing routine operating data to extract model parameter values and clustering them into reference clusters before actual mismatch detection. This preparation work is done using normal operational data, so when mismatch detection is needed, no additional experiments disrupting production are required - the detection can proceed immediately using the pre-established clusters, thus maintaining both detection capability and product quality.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentEP3296822B1Model-plant mismatch detection using model parameter data clustering for paper machines or other systems

Publication Date: 2022.02.16 HONEYWELL LTD(CA)

Figure 1
Figure 2~3
Figure 4

AI summary

A method includes repeatedly identifying (404) one or more values for one or more model parameters of at least one model (144, 230) associated with a process. The one or more values for the one or more model parameters are identified using data associated with the process. The method also includes clustering (406) the values of the one or more model parameters into one or more clusters (604). The method further includes identifying (408) one or more additional values for the one or more model parameters using additional data associated with the process. In addition, the method includes detecting (410) a mismatch between the at least one model and the process in response to determining that at least some of the one or more additional values fall outside of the one or more clusters. The values could be clustered using a support vector machine.