Forecasting the state of flow in rivers and canals, examining the flow passing over various surfaces and hydraulic structures, and monitoring the processes governing it due to various factors, has always been one of the most important researches of researchers. The passage of the current over the barrier, the effect of sea tides at the junction with the river and the waves caused by the dam failure upstream and downstream are examples of the mentioned hydraulic phenomena. In general, the flow in open channels is analyzed by Saint-Venant equations. In this research, the numerical solution of Saint-Venant's equations is discussed using deep learning and physics-informed neural networks (PINN) approach to simulate different phenomena in one-dimensional flow mode in open channels. In general, the numerical modeling of Saint-Venant's equations saves time, money and facilities compared to laboratory methods. This has long been the cause of expanding the attractiveness of numerical models and methods. In the present research, using deep learning and the physics-aware neural network (PINN) approach, a program is first set up in the Python software environment based on the aforementioned method and the SciANN software package. Then various problems including the one-dimensional and two-dimensional heat transfer problem in order to verify the results with the exact solution of the problem, the simple wave problem at the junction of the river with the sea along with the tide, the problems of supercritical and subcritical flow passing upstream of a ridge and The ideal dam failure problem with dry and wet downstream bed is investigated and the results are compared with the exact solution results. Finally, the failure of the laboratory dam (WES) on the sloping bed with bed friction and then the failure of the laboratory dam (CADAM) on the triangular ridge with the roughness of the bed downstream are investigated and the results are compared with the available laboratory data. The results obtained from the PINN method for one-dimensional and two-dimensional heat transfer problems show the accurate estimation of the method compared to the solution of Fourier analysis of these problems. Also, the obtained results indicate the favorable eva‎luation of the PINN method for the flow rate and water level of the stream in the problems of stable and gradually changing flows, including the simple wave problem, the passage of supercritical and subcritical flow upstream of a ridge (unblocked state) and the problem of dam failure. Ideal with dry and wet downstream bed as well as laboratory dam failure (WES) on sloping bed with bed friction. PINN method in eva‎luating flow rate and water level of fast changing streams, including problems of sub-critical flow in the upstream of an embankment with hydraulic jump and laboratory dam failure (CADAM) on a triangular embankment with bed roughness in the downstream, due to the discontinuity in the range Solving these issues, as well as drastic changes in depth and flow rate, do not provide optimal performance.

محمدنويد مقيم , بشير موحديان عطار

سعيد صرامي , احمد شانه ساز زاده