Embedded Controller Synthesis from Reduced System Identification Models

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current system identification methods are not suitable for automated deployment in embedded systems due to lack of guaranteed convergence and human interaction requirements, and they do not effectively combine iterative techniques with frequency-dependent model uncertainties, making them unsuitable for mass-produced products and satellite applications.

Innovation Solution

A method for automating system identification and controller synthesis that includes performing system identification experiments, fitting models to data, performing model reduction to generate suitable models for controller synthesis, and checking controller robustness to ensure stability and maximize closed-loop bandwidth and performance, using techniques like ARMAX models and frequency-dependent uncertainty estimation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If system identification is performed manually with engineer guidance, then model accuracy can be achieved, but the process requires human interaction and is not suitable for automated deployment

Engineering Contradiction:

Improvemodel accuracyVSAvoidautomated deployment capability

Core Design Contradiction:

Measurement precisionVSExtent of automation

Solution Approach 1:

The system performs self-identification by automatically collecting data, fitting models, evaluating uncertainties, and iterating without human intervention. The automated system identification process enables embedded systems to autonomously generate and validate their own dynamic models, eliminating the need for manual engineer guidance while maintaining model accuracy through iterative refinement.

Inventive Principle:

Principle #25Self-service

2Measurement precision

If iterative system identification techniques are used, then model accuracy improves, but convergence is not guaranteed and the process takes longer

Engineering Contradiction:

Improvemodel accuracyVSAvoididentification process time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system implements feedback by continuously evaluating frequency-dependent model uncertainties and using this information to guide iterative refinement. The uncertainty evaluation provides feedback on model quality, allowing the system to automatically adjust excitation signals and model parameters to improve accuracy while monitoring convergence criteria to terminate the process when sufficient accuracy is achieved.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The system changes parameters iteratively by adjusting model orders, excitation signal characteristics, and frequency ranges based on uncertainty evaluations. This parameter adaptation allows the system to converge to accurate models more efficiently by focusing computational resources on frequency regions with highest uncertainty.

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If complex models are used for system identification, then model accuracy improves, but the models are not numerically suitable for controller synthesis

Engineering Contradiction:

Improvemodel accuracyVSAvoidnumerical suitability for control

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The system segments the model development process into distinct phases: initial comprehensive model identification, uncertainty evaluation, and controlled simplification for control design. This segmentation allows the system to maintain high accuracy during identification while systematically reducing model complexity to levels suitable for numerical controller synthesis, preserving essential dynamics while removing unnecessary complexity.

Inventive Principle:

Principle #1Segmentation

4Productivity

If AI techniques are used for control, then performance may improve in specific cases, but reliability and trustworthiness cannot be guaranteed

Engineering Contradiction:

Improvecontrol performanceVSAvoidsystem reliability

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system replaces AI-based control approaches with physics-based dynamic models and classical control theory methods. By using fundamentally understood physical principles and mathematically rigorous control design techniques, the system achieves reliable and trustworthy control performance that can be guaranteed through proven theoretical frameworks rather than empirically tuned AI algorithms.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS11513479B1Automatic system identification and controller synthesis for embedded systems

Publication Date: 2022.11.29 AEROSPACE CORP

AI summary

A method for automating system identification includes performing a system identification experiment, and performing a system identifying processing by fitting a model to data from the system identification experiment. The method also includes performing model reduction to generate a model numerically suitable for controller synthesis by removing inconsequential states that cause controller optimization methods to fail. The method further includes performing control synthesis using the generated model or reduced models, including disturbance spectrum estimates, to generate a candidate controller design to be used during system operation. The method also includes checking for controller robustness using the identified model to ensure stability of the system while maximizing closed-loop bandwidth and performance.