Developed univariate regression and neural network multivariable models to predict the strength of welded joints in point values depending on their design parameters. It is shown that under conditions of multicollinearity factors one of the most effective methods to identify multivariate relationships between the mean values of the design parameters and the strength of welded connections is the point neural network modeling.
forecasting, strength, welded connection point, the regression model, the neural network model
"Prohnozyrovanye prochnosty svarnыkh tochechnыkh soedynenyi" [Prediction strength welded connection point],
Information Processing Systems,