Web-Based Predictive Modelling for Engineering Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing hybrid approaches for predicting the performance of complex engineering systems are complex and difficult for users to employ, as they require handling large parameter spaces and combining physics-based models with experimental data effectively.
Innovation Solution
A web-based system that includes a user interface for inputting engineering system properties and configurations, processing experimental data, generating operational profiles, training machine learning models using physics-based synthetic data, and predicting performance using a trained model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a hybrid approach combining physics-based models and experimental data is used for predicting engineering system performance, then prediction accuracy is improved, but system complexity increases
Solution Approach 1:
The patent introduces a web-based platform as an intermediary layer between the complex hybrid modeling system and the user. This platform handles data processing, model training, and prediction execution, shielding users from the underlying complexity while enabling access to high-accuracy predictions through standardized interfaces and automated workflows
Solution Approach 2:
The patent extracts the complex model training and data processing functions into a separate web-based service layer. By separating the computational complexity from the user interface, the system maintains high prediction accuracy through sophisticated hybrid modeling while presenting a simplified experience to users through the web platform
2Measurement precision
If a hybrid approach combining physics-based models and experimental data is used for predicting engineering system performance, then prediction accuracy is improved, but ease of operation deteriorates
Solution Approach 1:
The web-based platform implements self-service functionality by automatically handling data preprocessing, model selection, training, and validation. Users simply input their engineering system parameters and receive predictions without needing to understand or configure the complex hybrid modeling processes, making the high-accuracy tool accessible to non-experts
Solution Approach 2:
The web platform acts as an intermediary that translates simple user inputs into complex model execution. It manages the entire workflow from data ingestion to prediction generation, shielding users from operational complexity while delivering accurate results through automated model training and inference
3Measurement precision
If pure physics-based modelling is used, then theoretical accuracy is maintained, but ability to handle large parameter spaces deteriorates
Solution Approach 1:
The system performs preliminary actions by training machine learning models offline using extensive experimental data and physics-based simulations. These pre-trained models capture complex parameter relationships in advance, enabling the system to handle large parameter spaces efficiently during runtime while maintaining theoretical accuracy through physics-informed model architecture
Solution Approach 2:
The patent transforms the problem of handling large parameter spaces by using machine learning models to learn parameter relationships from data. The system changes from direct physics-based calculation for each parameter combination to using trained models that generalize across parameter spaces, maintaining accuracy while improving adaptability
Data Source
AI summary
The present invention discloses a system and method for web based performance predictive modelling of engineering system. The system provides a user interface for allowing a user to define at least one property and at least one configuration of an engineering system and to select a set of experimental data for predictive modelling. The system includes a computing device coupled in communication with the user interface is configured to receive the property and the configuration of an engineering system and the experimental data for predictive modelling. The computing device process the experimental data for creating a processed data set and also generate operational profiles for the engineering system. The computing device trains a machine learning model using the processed data and synthetic data to generate a trained machine learning model and predicts the performance of the engineering system using the trained machine learning model.