| 摘要 |
PURPOSE: A method for evaluating the performance of a tire is provided to improve the reliability of analysis about tire performance by preventing abnormal results from being saved in a system. CONSTITUTION: A method for evaluating the performance of a tire is as follows. Test data is scaled and data which is similar to data from a desired tire is extracted using a case-based reasoning Nearest Neighbor algorithm(S110). The difference between desired tire data and the extracted data is calculated(S120). Learning in an artificial neural network referring to the difference is initiated several times and number of hidden layers is obtained(S130). A predicted performance value is obtained by reflecting the tire data to the layer number of hidden layers(S140). Analysis is implemented in a boundary condition and an analysis condition after modeling a tire and composing a mesh. The analysis result is scaled and compared with the predicted performance value to determine whether the analysis result exists in a set reliability range(S150).
|
