Technical System Configuration Using Reference Model Transfer

Solution Overview

Problem

Existing methods for optimizing process parameters in production and machining processes require large quantities of measurement data, leading to high time and cost expenditures.

Innovation Solution

A method that involves detecting reference observations from technical reference systems, conditioning relevant models for these systems, and using these models to adjust and configure a technical system to be optimized, thereby reducing the need for extensive measurement data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If model-based optimization methods are used to ascertain process parameters, then desired properties of workpieces can be achieved, but large quantities of measurement data are required leading to high time and cost expenditures

Engineering Contradiction:

Improvedesired properties of workpieceVSAvoidtime expenditure

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent applies preliminary action by pre-training a source model on a source domain with available measurement data before optimizing the target system. This pre-learned model serves as a starting point that already captures relevant process-parameter relationships, eliminating the need to start from scratch and reducing the amount of new measurement data required for target system optimization.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent implements copying by creating a target model that is initialized as a copy or adaptation of the pre-trained source model. The target model inherits the knowledge and parameters learned from the source domain, allowing the optimization process to begin with a knowledgeable baseline rather than requiring complete re-learning from scratch.

Inventive Principle:

Principle #26Copying

2Manufacturing precision

If model-based optimization methods are used to ascertain process parameters, then desired properties of workpieces can be achieved, but large quantities of measurement data are required leading to high cost expenditures

Engineering Contradiction:

Improvedesired properties of workpieceVSAvoidquantity of measurement data

Core Design Contradiction:

Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent applies preliminary action by pre-training a source model on a source domain with available measurement data before optimizing the target system. This pre-learned model serves as a starting point that already captures relevant process-parameter relationships, eliminating the need to start from scratch and reducing the amount of new measurement data required for target system optimization.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent implements copying by creating a target model that is initialized as a copy or adaptation of the pre-trained source model. The target model inherits the knowledge and parameters learned from the source domain, allowing the optimization process to begin with a knowledgeable baseline rather than requiring complete re-learning from scratch.

Inventive Principle:

Principle #26Copying

3Productivity

If transfer learning is used with a model already learned to reduce measurement data requirements, then data efficiency improves, but the complexity of conditioning and adjusting models increases

Engineering Contradiction:

Improvedata efficiencyVSAvoidmodel conditioning complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent applies segmentation by separating the model adaptation process into distinct phases: first conditioning the source model on source domain observations, then adjusting the conditioned model to fit target domain observations. This segmentation of the complex transfer learning process into manageable stages reduces the perceived complexity while maintaining data efficiency benefits.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent implements feedback by using the conditioned source model to generate predictions on target domain data, then adjusting the model based on the difference between predictions and actual target observations. This iterative feedback mechanism systematically reduces the complexity of model adjustment by providing clear guidance on what modifications are needed.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS20250077726A1Method for configuring a technical system to be configured

Publication Date: 2025.03.06 ROBERT BOSCH GMBH

AI summary

A method for configuring a technical system. The method includes: detecting reference observations; conditioning reference system models on the reference observations detected for the reference system; detecting observations of results of the technical system to be configured for different values of the configuration parameters; adjusting an a priori model for the relationship between the values of the configuration parameters and the results provided by the technical system to the observations detected for the technical system, wherein the a priori model is formed from a weighted combination of the conditioned reference system models; ascertaining an a posteriori model for the relationship between the values of the configuration parameters and the results provided by the technical system by conditioning the adjusted a priori model on the observations detected for the technical system to be configured; and configuring the technical system using the ascertained a posteriori model.