Based on the measurement data on the sediment concentration of section AD4 in the north passage of south branch of the Yangtze estuary on April 26, 2009, interpolation calculation of sectional sediment concentration distribution was conducted using inverse distance to a power method, modified Shepard's interpolation, polynomial regression interpolation method, Kriging interpolation method and radial basis function interpolation method provided by SURFER 8 0 software, and the calculated values were compared with the actual measured values As indicated by the measurement result of interpolation accuracy according to the calculation results of absolute error, relative error, root-mean-square error and goodness value of prediction, the Kriging interpolation method gives the best interpolation accuracy of calculation of sectional sediment concentration distribution Based on the result this, interpolation calculation and comparison were carried out with BP neural network, radial basic function neural network and generalized regression neural network It is discovered that generalized regression neural network is characterized by high interpolation accuracy, fast convergence speed and convenient operation, being an effective interpolation method Interpolation calculation of sediment concentration distribution with neural network can simulate the complicated nonlinear relationship between the sediment concentration and section spatial coordinates, and perform nonlinear optimization In interpolation with-neural network, the interpolation points may be free from the limitation of the output form of other interpolation mesh points, thus facilitating research and application
