| Summary: | 碩士 === 國立臺灣師範大學 === 工業教育研究所 === 86 === The curves of tool path are the main concerns in the machining processes. How to get the optimal curve passing through those desired points will be the goal of this research.
In this study the Genetic Algorithm with the multi-pathed searching is used to find the optimal solution, and compared with the results from the single-pathed NUOF Transform.
Traditionally, the Genetic Algorithm is used for discrete datum. Because the parameters of the NUOF Series are real number, the algorithm for real-valued datum is used. It sounds reasonable to compute the coefficients of NUOF by a real-valued Genetic Algorithm. To do so, the error due to transformation from continuous to discrete datum can be avoided and vice versa. The computing time and memory space are lessened because of no need for coding and decoding. As the initial population are generated randomly, the answers obtained from processing reproduction, crossover and mutation are much different due to different number of generations. Finally, we found that the frequency in different frequency domains would converge in the vicinity of certain frequency. We can say this frequency is the optimal frequency in that frequency domain.
|
|---|
