Industrial Process Model Training With Selective Simulation Data

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Industrial applications of machine learning face challenges due to the scarcity of labeled training data, with high-fidelity simulations being time-consuming and computationally expensive, and blind data generation lacking guaranteed impact on algorithm performance.

Innovation Solution

An industrial process model generation system that utilizes a simulator to generate behavioral data, which is then used to train machine learning algorithms, with the system deciding which data subsets to use for training based on performance conditions, optimizing computational efficiency and model improvement.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If high-fidelity simulations are used to generate training data, then model training data availability is improved, but computational cost and time consumption increase

Engineering Contradiction:
Improvetraining data availabilityVSAvoidcomputational cost
Core Design Contradiction:
Quantity of substanceVSUse of energy by stationary object

Solution Approach 1:

The system performs partial simulations by selectively executing simulations only for specific operating conditions or scenarios that are most valuable for training, rather than comprehensively simulating all possible conditions. This reduces computational cost while still generating sufficient training data for model development

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The system pre-generates training data through simulations for anticipated operating conditions before actual model training is needed. This preliminary data generation allows for more efficient model training later without requiring extensive real-time computational resources

Inventive Principle:
Principle #10Preliminary action

2Quantity of substance

If blind data generation is performed, then data quantity is increased, but data quality and impact on algorithm performance are not guaranteed

Engineering Contradiction:
Improvedata quantityVSAvoiddata quality
Core Design Contradiction:
Quantity of substanceVSReliability

Solution Approach 1:

The system implements feedback mechanisms where simulation results are evaluated against performance metrics, and subsequent simulations are adjusted based on this feedback. This ensures that generated data actually improves model performance rather than merely increasing data quantity

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The system dynamically adjusts simulation parameters based on model performance requirements and data distribution analysis. By changing parameters such as operating conditions, disturbance levels, and scenario types, the system generates diverse high-quality data that针对性地 addresses model training needs

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3916496B1An industrial process model generation system
Publication Date: 2025.01.01 ABB (SCHWEIZ) AG
  • EP3916496B1 patent drawingFigure 1
  • EP3916496B1 patent drawingFigure 2
  • EP3916496B1 patent drawingFigure 3

AI summary

The invention relates to an industrial process model generation system, comprising: an input unit; and a processing unit. Tthe input unit is configured to receive a plurality of input value trajectories comprising operational input value trajectories and simulation input value trajectories relating to an industrial process. The processing unit is configured to implement a simulator of the industrial process. The processing unit is configured to generate a plurality of industrial process behavioural data, wherein industrial process behavioural data is generated for at least some of the plurality of input value trajectories, and wherein the generation of the industrial process behavioural data for the at least some of the plurality of input value trajectories comprises utilization of the simulator. The processing unit is configured to implement a machine learning algorithm that models the industrial process. The processing unit is configured to train the machine learning algorithm. The processing unit is configured to process a first behavioural data of the plurality of behavioural data with the machine learning algorithm to determine a first modelled result. The processing unit is configured to determine to train or not to train the machine learning algorithm using the first behavioural data, the determination comprising a comparison of the first modelled result with a performance condition. The processing unit is configured to process a second behavioural data of the plurality of behavioural data with the machine learning algorithm to determine a second modelled result. The processing unit is configured to determine to train or not to train the machine learning algorithm using the second behavioural data or to further train or not to further train the machine learning algorithm using the second behavioural data, the determination comprising a comparison of the second modelled result with the performance condition