Hybrid Synthetic Machine Data Validation for Fault Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Physics-based models for sensor data processing are time-consuming due to reliance on physical simulations, while generative learning, though faster, struggles with near-real-time data generation accuracy.
Innovation Solution
A hybrid approach combining Finite Element Analysis (FEA) with a variational autoencoder-based generative model for end-to-end data synthesization and validation, incorporating physical characteristics into the latent representation and using a dual validation framework involving rule engine and AI-based validation techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If physics-based models are used for sensor data processing, then data quality is improved, but processing time increases
Solution Approach 1:
The patent pre-computes and stores physics-based simulation data for various fault conditions before actual diagnosis is needed. This offline preparation creates a library of synthetic training data that captures complex physics relationships, eliminating the need for real-time physics simulations during actual fault diagnosis operations.
Solution Approach 2:
The patent creates synthetic copies of real sensor data through physics-based simulations and generative models. These synthetic data copies replicate the statistical properties and fault patterns of real data, enabling training of diagnostic models without requiring actual physical simulations during operation.
2Productivity
If generative learning is used for data generation, then processing speed is improved, but data quality deteriorates
Solution Approach 1:
The patent introduces physics-based simulation data as an intermediary between real sensor data and the generative model. The generative model is trained on physics-simulated data that has been carefully constructed to preserve realistic fault patterns, thereby mediating between the speed advantages of generative models and the quality requirements of diagnostic applications.
Solution Approach 2:
The patent modifies the training parameters and architecture of the generative model to prioritize preservation of fault-related statistical properties over mere data generation speed. By adjusting parameters such as noise levels, generation steps, and loss function weights, the system optimizes for diagnostic quality while maintaining computational efficiency.
3Measurement precision
If physics simulations are performed in real-time, then data accuracy is improved, but system complexity increases
Solution Approach 1:
The patent performs complex physics simulations offline before deployment, pre-computing synthetic data for various fault scenarios. This preliminary action transfers the computational burden from runtime to offline preparation, simplifying the deployed system while maintaining high data accuracy through physics-based generation.
Solution Approach 2:
The patent replaces complex real-time physics simulations with pre-generated synthetic data copies. These copies capture the essential physics relationships without requiring complex simulation engines during operation, thereby reducing system complexity while preserving data accuracy.
Data Source
AI summary
In sensor data analytics, physics-based models generate high quality data. However, these models consume lot of time as they rely on physical simulations. On the other hand, generative learning takes much less time to generate data, and may be prone to error. Present disclosure provides system and method for generation of synthetic machine data for healthy and abnormal condition using hybrid of physics based and generative model-based approach. Finite Element Analysis (FEA) is used for simulating healthy and faulty parts in machinery with set of parameters and pre-condition(s). Small output data from FEA is fed into a generative model for generating synthesized data by learning data distribution knowledge and representing into latent space. Rule engine is built using statistical features wherein realistic bounds serve as faulty data indicators. Synthesized data which does not satisfies features bounds are discarded. Further, AI-based validation framework is used to analyze quality of synthesized data.