Fremgangsmåte for å danne en optimalisert neural nettverksmodul tilsiktet å simulere strömningsmodus for enflerfaset fluidström

摘要

- method for constructing a module (or hydrodynamic thermodynamic for example) intended to simulate a real time the flow mode at any point of a pipe, of a stream of the multiphase fluid comprising at least one liquid phase and at least one gas phase, so that it is best suited to the operating conditions are fixed on a certain number of structural parameters and physical defined relative to the pipe, and on a set of physical magnitudes defined (hydrodynamic or thermodynamic quantities, for example), with the range of variation are fixed to the parameters and the physical magnitudes. - it comprises the use of a system of simulation based on neural networks is not linear, with each of the inputs for the parameters of the structure and the physical magnitudes, and outputs which are available of the quantities necessary for the estimation of the flow mode, and at least one intermediate layer. The neural networks are determined iteratively to be adjusted to the values of a learning base with the predefined tables connecting different values obtained for the output data to the corresponding values of the input data. It uses a learning base adapted to the operating conditions imposed and it generates of the neural networks optimized fitting, at best, the operating conditions imposed. - applications in the modelling of flow of hydrocarbons in the pipes, for example.