Sash sheet metal development and the application of a back-propagation neural network in the calculation of the developed length

All of the sides constituting a sash sheet metal are most squares or rectangles, only the sections of sheet metal needing to be designed. The rest of the entire sash sheet metal product can be generated from the width of the sash sheet metal entered by the user, and the section shape of the sheet metal designed by the user. Before a sheet metal product is developed, the developed length needs to be calculated. The calculation of the developed length requires the curve data for the 'relative curve drawing of the bending radius, the sheet metal strip thickness and the correction factor'. This paper uses a back-propagation neural network to search for the relationship between the two coordinate axis data of the curve. This relationship is stored in the connected weight amongst neurons. Therefore, it is possible to simply run the attribute value data to be inferred through the weight matrix calculation to quickly obtain the network inference results, which also possess the capability of uncertainty inference. By means of the weight matrix, the correction factor can be obtained from the 'bending radius, sheet metal strip thickness and correction factor relative curve drawing' to calculate the developed length. (C) 1997 Elsevier Science S.A.