Reference solutions + trained models

The data are used to reproduce the results from the paper "Physics-informed neural networks for one-dimensional sound field predictions with parameterized sources and impedance boundaries" by Borrel-Jensen et al. Realistic sound is essential in virtual environments, such as computer games and mixed reality. Efficient and accurate numerical methods for pre-calculating acoustics have been developed over the last decade; however, pre-calculating acoustics makes handling dynamic scenes with moving sources challenging, requiring intractable memory storage. A physics-informed neural network (PINN) method in 1D is presented, which learns a compact and efficient surrogate model with parameterized moving Gaussian sources and impedance boundaries, and satisfies a system of coupled equations. The model shows relative mean errors below 2%/0.2 dB and proposes a first step in developing PINNs for realistic 3D scenes. ----------- The data contains: * Reference solutions in HDF5 format for validating the predictions * Trained models used in the result section of the paper Code is available here: https://github.com/dtu-act/pinn-acoustic-wave-prop本数据集用于复现Borrel-Jensen等人发表的论文《带有参数化声源与阻抗边界的一维声场预测物理知情神经网络》（Physics-informed neural networks for one-dimensional sound field predictions with parameterized sources and impedance boundaries）中的实验结果。 虚拟现实环境（如电脑游戏与混合现实）中，逼真的音效至关重要。近十年来，学界已开发出高效精准的预计算声学数值方法，但此类预计算方法难以处理包含移动声源的动态场景，会面临难以承受的内存存储开销。 本文提出一种一维物理知情神经网络（Physics-informed neural networks, PINN）方法，该方法可学习得到兼具紧凑性与高效性的代理模型，支持参数化的移动高斯声源与阻抗边界，并满足耦合方程组约束。该模型的相对平均误差低于2%/0.2分贝，为面向真实三维场景的物理知情神经网络开发迈出了第一步。 本数据集包含： * 用于验证模型预测结果的HDF5格式参考解 * 论文结果章节中使用的已训练模型 代码开源地址：https://github.com/dtu-act/pinn-acoustic-wave-prop