Code and Datasets for Parameter Identification of Thermoacoustic Oscillations using PI-DeepONet

This repository contains the source code, numerical training data, and validation benchmarks associated with the study on "Parameter Identification of Thermoacoustic Oscillations using PI-DeepONet."The dataset includes:Source Code: PyTorch implementation of the PI-DeepONet architecture, physics-informed constraints, and the training engine.（The v2 version provides a PI-DeepONet code framework for jointly training first- and second-order moments, as well as code for generating simulated data.）Training Data: High-fidelity numerical solutions of the Fokker-Planck equations (Ground Truth) used for supervising the network, covering both amplitude and displacement increment expectations.Validation Data: Synthetic Stochastic Differential Equation (SDE) time-series generated via Latin Hypercube Sampling (LHS).This dataset is intended for reproducing the reported results and benchmarking parameter inversion algorithms for stochastic dynamical systems.