Model Predictive Controller Updating via White Noise Diagnostics

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Model Plant Mismatch (MPM) leads to inaccurate predictions and performance degradation in Model Predictive Control (MPC), requiring lengthy and expertise-dependent model re-identification processes that result in production of low-quality products during perturbation periods.

Innovation Solution

A method and system for updating a model predictive controller by assessing performance deviations, diagnosing MPM through white noise addition to set points, calculating model prediction errors, and quantifying mismatch using correlation coefficients, thereby reducing the perturbation period and efficiently updating the model.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional model re-identification is performed to diagnose MPM, then model accuracy is improved, but perturbation period increases and productivity decreases

Engineering Contradiction:
Improvemodel accuracyVSAvoidproduct quality
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent changes the parameters of the perturbation signal by using optimized white noise characteristics and adjusting the excitation level to achieve sufficient model identification accuracy with minimal perturbation duration, thereby reducing off-spec product generation

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The system performs preliminary diagnostics using process data before initiating full model re-identification, allowing early detection of MPM conditions and preparation of updated models in advance to minimize perturbation periods

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If traditional model re-identification is performed to diagnose MPM, then model accuracy is improved, but time consumption increases

Engineering Contradiction:
Improvemodel accuracyVSAvoidperturbation period
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent implements periodic model updating at predetermined time intervals or when performance degradation thresholds are reached, allowing the system to maintain acceptable model accuracy without continuous perturbation, thus reducing total perturbation time

Inventive Principle:
Principle #19Periodic action

Solution Approach 2:

The system applies partial model updating by re-identifying only specific model parameters or subsystems that have degraded rather than performing complete model re-identification, significantly reducing the perturbation period required

Inventive Principle:
Principle #16Partial or excessive action

3Ease of operation

If automated model updating is implemented, then ease of operation is improved, but device complexity increases

Engineering Contradiction:
Improvemodel updating automationVSAvoidsystem complexity
Core Design Contradiction:
Ease of operationVSDevice complexity

Solution Approach 1:

The system performs self-diagnosis and self-updating of models by automatically detecting MPM conditions, selecting appropriate identification methods, and implementing model updates without operator intervention, improving ease of operation while managing complexity through automation

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP2561411B1A method and system for updating a model in a model predictive controller
Publication Date: 2020.11.04 ABB (SCHWEIZ) AG
  • EP2561411B1 patent drawingFigure 1
  • EP2561411B1 patent drawingFigure 2
  • EP2561411B1 patent drawing

AI summary

The invention relate to a method for updating a model in a model predictive controller. The method comprises assessing the deviation of the operating performance level from the desired performance level of the process plant. Diagnosing the model predictive control for the model plant mismatch by updating the model in a model predictive controller. Diagnosing the model predictive controller comprises determining the model prediction error in relation to model plant mismatch, quantifying the model plant mismatch, and updating the model in the said model predictive controller. The invention also relate to a system for updating a model in a model predictive controller in accordance with the method of the invention.