Aero-engine Dynamic Pressure Calibration Method with Deep Learning and Wavelet Transform

To measure the high-temperature dynamic pressure of aircraft engines, a pressure probe device is often used in view of the temperature resistance and economy of the sensor. However, such a device will affect the dynamic characteristics of the pressure sensor and cause distortion of the measurement signals. Therefore, this paper proposes an intelligent calibration method for dynamic pressure signals, which extracts frequency information of dynamic pressure signals through synchrosqueezing wavelet transform (SWT) and uses a Bidirectional Long Short-Term Memory (BiLSTM) neural network to calibrate the distorted signals. The results show that the average relative error between the calibrated dynamic pressure signals of the model and the standard dynamic pressure signals is within 2.5%, indicating a high calibration accuracy.