The synchronous belt optimum form forecasting device and synchronous belt optimum form forecasting manner, and

摘要

PPROBLEM TO BE SOLVED: To predict an optimum shape of a toothed belt in a short time and with high accuracy. PSOLUTION: A plurality of dimensional parameter groups are formed by selecting a plurality of dimensional parameters composing a toothed belt, setting an actual value of each parameter and a value therearound as a plurality of level values with respect to a current toothed belt and arbitrarily combining each level value of each parameter (S1-3). After obtaining output which indicates performance of the toothed belt through the FEM analysis by taking as input the preset physical property data and mechanical condition of the toothed belt and a plurality of dimensional parameters, the neural network is made to learn the combination of the input and output (S4-5). A plurality of identification data are formed by arbitrarily setting a value of each dimensional parameter in a range between the maximum value and minimum value of a level value and input in a learnt neural network to obtain identification output, thereby determining the optimum shape of a toothed belt satisfying specified constraint conditions and having the recognition data to which the best recognition output is provided (S6-8). PCOPYRIGHT: (C)2007,JPO&INPIT