A pinn-based sound field visualization acquisition system and method

Abstract

The application discloses a sound field visualization collection system and method based on PINN. After an ultrasonic probe receives a signal generated by a signal generator, the signal is transmitted to a collection area, a hydrophone collects a sound pressure signal and then sends the sound pressure signal to an oscilloscope, and the oscilloscope transmits the sound pressure signal to an industrial computer. The method comprises the following steps: constructing a physical model partial differential equation and a PINN model by using sample data, establishing a constraint condition for the PINN model by the partial differential equation and determining a loss function Loss, training the PINN model by using the sample data, inputting data of a to-be-measured collection point into the trained PINN model, obtaining a change condition of a sound field with time, and realizing visualization of a propagation process of sound waves in the sound field. The application solves the problem that a large amount of collection point data is required for sound field visualization, effectively reduces the time cost of sound field visualization, and can quickly restore the propagation process of the sound field and realize visualization by using less collection point data.

Images