VTOL Air Data Calibration Using Neural Networks Near the Fuselage

The patent replaces the traditional mechanical pitot-static measurement system with an artificial neural network-based computational system. The neural network processes multiple sensor inputs (pitot pressure, static pressure, temperature, accelerometer data, gyroscope data) to compensate for the effects of air flows and turbulences, eliminating the need to place sensors in locations optimized for mechanical measurement accuracy.

The patent creates a multi-functional air data indicating arrangement where a single integrated system performs multiple functions: it processes pressure data from pitot and static ports, temperature data, acceleration data, and rotational data simultaneously through the neural network to provide accurate air speed and altitude measurements regardless of sensor placement or flight conditions.

The patent performs preliminary training of the artificial neural network using comprehensive flight test data covering various flight conditions, configurations, and maneuvers before actual operation. This pre-training phase establishes the neural network's ability to accurately process sensor data across all expected operating scenarios, eliminating the need for time-consuming iterative calibrations during the design process.

The system incorporates feedback mechanisms where the neural network continuously processes real-time sensor data and adjusts its outputs based on patterns learned during training. The system can also incorporate actual flight data to refine and update the neural network models, enabling continuous improvement without requiring complete recalibration.