Physics-informed neural networks (PINNs) often struggle to balance multiple loss terms in thermally coupled multiphysics problems. We propose Cooperative Soft-Hard PINNs (s-hPINN/s-HB-PINN), which apply soft constraints to fields with Neumann conditions while enforcing hard constraints on others to balance exact boundary enforcement with training stability. Validated on thermoelasticity and thermal convection, our method reduces training time by approximately 56%. In thermal convection experiments, incorporating partial data further reduces velocity errors by up to 78% compared to standard PINNs. We subsequently assessed the framework’s robustness against varying relative Gaussian white noise levels and different data sampling locations. The result demonstrate that s-HB-PINN maintains high-fidelity predictions even under noise interference, consistently outperforming baseline methods. This confirms that the proposed collaborative strategy offers a superior trade-off between accuracy, efficiency, and robustness in complex multiphysics environments.

As a mesh-free solving paradigm, Physics-Informed Neural Networks (PINNs) demonstrate potential in both forward and inverse problems by embedding physical equations into the loss function. However, they still face challenges in capturing the spatiotemporal evolution of complex physical processes. When applied to time-dependent complex flows, such as high-Reynolds-number cylinder flow, they often rely on supervised data, which is frequently difficult to obtain accurately in practice. To address these issues, this paper proposes a novel unsupervised solving framework—the Adaptive Hard-Constraint Physics-Informed Neural Network (AHC-PINN). This method integrates an adaptive sampling mechanism based on partial differential equation residuals with a hard-constraint strategy. By dynamically evaluating the contribution of collocation points to the loss and incorporating analytically embedded boundary constraints, it directs the network training entirely toward solving the governing equations. Using two-dimensional unsteady cylinder flow as a validation case, experimental results show that AHC-PINN significantly improves the prediction accuracy of wake evolution under unsupervised conditions. Its performance surpasses that of traditional soft-constraint PINNs by an order of magnitude and is even superior to methods using sparse supervised data. Furthermore, through analysis of the PDE loss and gradient distribution, the study explicitly identifies the impact of large-gradient regions on PINN training stability and prediction accuracy, providing a basis for subsequent optimization.