Research on Simulation and Dynamic Pressure Prediction of Pipeline Transient Conditions Based on Fluent and Deep Learning

The study of pressure change characteristics under failure conditions of pipeline sections, such as pipe burst and leakage loss, is of great significance to the safety of urban water supply network and leakage prevention and control. At present, most of the related technical research is based on the steady-state hydraulic model, which has low accuracy and cannot accurately obtain the transient law of water flow in the pipeline under the failure conditions. To solve this problem, a transient hydraulic model of pipeline failure is established based on the Navier-Stokes equations, the finite volume method and the hydrodynamic control equations, which is solved by the finite element analysis method to realize the pipeline transient pressure simulation and transient characteristic analysis. On this basis, based on the transient pressure data, the pipeline transient pressure simulation and prediction model based on the LSTM neural network is constructed to analyze the pressure response characteristics of the pipeline under the failure conditions, and to dynamically predict the transient pressure of the pipeline at multiple points during the occurrence of bursting, leakage, and so on. Example applications show that the average absolute percentage error of pipeline transient pressure prediction is stabilizes at less than 0.2%, and the constructed pipeline failure transient hydraulic model and transient pressure simulation model have high computational accuracy. The coupled application of the two models can realize the dynamic simulation and prediction of the pressure under the pipeline failure conditions, which can help to provide an important theoretical basis for the further development of scientific and efficient leakage state pattern identification and accurate leakage location through the transient pressure characterization, and provide effective technical support for water supply enterprises in leakage monitoring and control.