Calculation of Photoionization Rates During Streamer Discharge Using Neural Networks

Photoionization is a significant source of electrons in streamer discharge. Typically, the photoionization rate is calculated using the integral model or three-exponential Helmholtz model in streamer discharge simulations. However, these models exhibit certain disadvantages, such as low calculation speed and accuracy. To overcome these drawbacks, this article proposes a novel method for calculating the photoionization rate using neural networks (NNs). Two types of datasets were used to train the NNs, namely, the random and Gaussian datasets. The results indicate that the performance of the model when trained using Gaussian datasets is better than that when random datasets are used. Furthermore, the NN model exhibits high computational accuracy under the test conditions, verifying its use as a promising technique to calculate photoionization rates. Additionally, the effects of hyperparameters and attention mechanisms are evaluated, laying the groundwork for the application of NNs in practical streamer models.