Closed-Loop Control Function Selection Using Counterfactual Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for changing technical systems or processes lack robustness and explainability, particularly in scenarios with imperfect causal knowledge, leading to inefficiencies and difficulties in predicting the effects of changes.
Innovation Solution
A method and device that utilize models to analyze and select functions for change based on learned distributions and noise posteriors, enabling counterfactual analysis to predict the effects of changes, thereby improving the likelihood of achieving desired outcomes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current methods are used to analyze and improve technical systems, then system changes can be implemented, but the analysis lacks robustness and explainability under imperfect causal knowledge
Solution Approach 1:
The patent applies preliminary action by predicting the effects of potential system changes before actually implementing them. The method uses learned causal models to simulate and evaluate multiple possible changes, selecting the optimal change based on predicted outcomes. This allows developers to make informed decisions about system modifications without having to actually implement and test each change in the real system, thereby providing robust and explainable analysis while avoiding the risks of trial-and-error modifications.
2Productivity
If system changes are implemented directly, then improved performance can be achieved, but the risks and costs of actual system modifications increase
Solution Approach 1:
The patent applies copying by creating and using learned causal models that replicate the behavior of the actual technical system. These models serve as virtual copies that can be modified and tested without affecting the real system. The method trains causal models on historical data to capture the system's behavior, then uses these models to predict the effects of potential changes. This allows performance improvement to be explored safely in the virtual model before any actual system modification is made, eliminating the risks associated with direct system changes.
3Reliability
If multiple potential changes are evaluated, then the likelihood of finding an effective improvement increases, but the computational complexity and time required increase
Solution Approach 1:
The patent applies preliminary action by pre-training causal models on historical system data before needing to evaluate changes. This training phase captures the system's causal relationships and behavior patterns in advance. When it's time to evaluate potential improvements, the pre-trained models can quickly predict the effects of multiple candidate changes without requiring extensive computational resources or time. This preliminary preparation enables rapid assessment of multiple changes while maintaining high reliability in the predictions.
Solution Approach 2:
The patent applies mechanics substitution by replacing actual system modifications with virtual simulations using learned causal models. Instead of physically or operationally implementing changes in the real system to evaluate them, the method uses computational models to simulate change effects. This substitution dramatically reduces the time and resources required for change assessment while maintaining accurate predictions, as the virtual simulations can be executed rapidly without affecting the actual system operation.
Data Source
AI summary
A device and method for analyzing a technical system or process that includes a first function and a second function. A first model models a functional relationship of input and output variables of the first function and an influence of a first disturbance variable. The second model models a functional relationship of input and output variables of the second function and an influence of a second disturbance variable. An observation is provided that includes input variables and output variables of the first function and of the second function. A prediction regarding an effect of changing the first function under the influence of the first disturbance variable and/or changing the second function under the influence of the second disturbance variable is determined, and of these functions, that function whose change has a higher likelihood of achieving a desired effect is selected.