Microwave Sensor AI Model Training for Measurement Accuracy

The patent replaces traditional analytical measurement models with artificial intelligence-based models (neural networks, support vector machines, random forests) that can capture complex non-linear relationships between sensor data and measurement variables. This substitution enables higher measurement accuracy without requiring explicit analytical formulations, as the AI models learn patterns from training data and generalize to new measurements.

The patent transforms the measurement approach by changing from using fixed analytical equations to using learned parameters from training data. The AI models are trained on labeled data containing sensor measurements and corresponding ground truth values, allowing the system to adapt to complex relationships by adjusting model parameters during training rather than relying on pre-defined physical equations.

The patent applies preliminary action by performing model training in advance using training data that is collected and processed before actual measurements are taken. During training, the AI model learns the complex relationships between sensor data and measurement variables, and this learned knowledge is stored in the model parameters. During actual measurement, the system only needs to input sensor data into the trained model, avoiding the need for real-time complex computations and significantly reducing measurement cycle time.

The patent uses copying by creating a simplified representation of the complex physical process through the AI model. The model captures the essential relationships between sensor data and measurement variables by learning from training data, creating a computational copy that can be evaluated quickly without re-simulating the entire physical process. This copy allows for fast predictions while maintaining accuracy.