Closed-Loop Test Platform Validation Using Similarity Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in virtual validation or verification of closed-loop test platforms is the inadequate simulation of the system's surroundings, leading to potential operational reliability issues, especially in complex systems where feedback loops are involved, as existing methods may not accurately represent real-world scenarios.
Innovation Solution
A method involving a machine learning module trained on data sets from reference test platforms to estimate the similarity of input and output data, allowing for reliable validation or verification of closed-loop test platforms by comparing and classifying the behavior of the system under test.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If virtual validation or verification is used to reduce testing scope and resources, then productivity and cost efficiency are improved, but the reliability of validation results deteriorates due to inadequate simulation of real-world scenarios
Solution Approach 1:
The patent implements feedback by comparing simulation results with real-world test data and using the discrepancies to iteratively improve the simulation model. The system continuously refines the virtual environment based on feedback from actual system behavior, ensuring that the simulation accurately represents real-world scenarios while maintaining high validation efficiency.
Solution Approach 2:
The patent applies preliminary action by pre-training machine learning models with extensive real-world data before using them to validate simulation results. This preliminary preparation ensures that the validation process can quickly and reliably assess whether virtual test scenarios accurately represent real-world conditions, maintaining both efficiency and reliability.
2Adaptability or versatility
If complex systems are simulated in closed-loop test platforms to test feedback behaviors, then the comprehensiveness of testing is improved, but the device complexity increases making validation more difficult
Solution Approach 1:
The patent introduces machine learning models as intermediaries that bridge the complex simulation environment and the validation process. These models learn to recognize patterns and behaviors in the complex system outputs, translating complex simulation data into meaningful validation metrics without requiring direct complex analysis of all system interactions.
Solution Approach 2:
The patent creates simplified digital twins or representative models of complex subsystems that capture essential behaviors without replicating full complexity. These copied models enable comprehensive testing of feedback behaviors while reducing the overall platform complexity and making validation more manageable.
3Ease of operation
If synthesized input signals are used in simulation to represent real surroundings, then the ease of operation is improved, but the measurement precision deteriorates as relevant features may be changed or lost
Solution Approach 1:
The patent dynamically adjusts parameters of synthesized input signals based on learned characteristics from real-world data. The machine learning models identify which parameters are critical for system behavior and modify only those parameters in synthesized signals, maintaining signal fidelity for relevant features while preserving the ease of generating synthetic test inputs.
Solution Approach 2:
The system employs self-service by using the synthesized signals themselves to train the machine learning models that subsequently validate them. The models learn to recognize when synthesized signals adequately represent real-world conditions, enabling the system to self-validate signal fidelity without external intervention.
Data Source
AI summary
A method for training a machine learning module to validate or verify a closed-loop test platform for testing a system. The method includes: providing data sets from multiple closed-loop test platforms of which at least one is a reference test platform, each data set containing input data and associated output data of a system under test in a respective closed-loop test platform; determining the similarity of portions of input data for pairs of the data sets, each pair of test data sets containing a data set of a reference test platform; determining similarities of portions of the output data associated with the portions of input data; and training a machine learning module to estimate the similarity.