自行車變速系統測試之研究

Abstract

在自行車中，變速系統提供了不同之速比組合用來適合多變之騎乘環境，騎乘者可藉由調整變速撥桿選擇快速或省力之騎乘狀況。因此，變速性能之良窳直接影響到騎乘者之感覺，對設計者而言是十分重要之考慮因素。本文主要針對變速系統提出一系統化之測試流程。在變速系統上，應用"穩定區"之觀念及系統整合之技術可使變速系統零組件運轉順暢及更有效率，故本文對穩定區之理論、特性、實驗方法及其應用作詳細的介紹。在變速系統齒盤上的測試，主要以上鏈型式之判別與性能測試為主。性能測試為變速系統於換檔過程中其變速功能之判定，經由性能測試輔助以上鏈型式之判別，便可得知在齒盤上，輔助上、下鏈點設計或製造上之缺陷並加以改善，另外也可針對不同廠牌之飛輪其變速性能加以比較，如此便可以取代傳統以騎乘者實車測試，其標準將更為客觀。另外，為了讓測試變得更有效率及準確，所以在本文中將使用最佳化的工具如田口方法及實驗設計法，及應用人工智慧中之類神經網路及模糊理論來達成。

The derailleur system on a bicycle serves the same purpose as a transmission in a motor vehicle: it provides different speed ratios for different situations. The cyclists can choose either speed or labor-saving by moving shift-levers to achieve a series of mechanical motions for changing gears. Therefore, the shifting performance of the derailleur system is very important in designing the components of the derailleur system. This thesis proposes a systematic procedure for arranging the testing steps for the derailleur system. In system testing, the "stability region" concept and the system integration technique enable that the components of the derailleur system can integrate well and operate effectively. Stability region properties are introduced and procedures for the stability region test are also proposed. Then, the concepts and experimental results of the stability region can be used in the component selection and integration of a derailleur system. In testing the freewheel or chainwheel sprocket, the classification of chain engagement types and performance tests are two major interests. The results of performance tests can be used to judge the shifting performance during shifting gears. In addition, base on the results of performance tests, some design and manufacturing imperfections of products can also be found. The experimental procedures and results provide the designer a more objective design criteria instead of using the subjective human "feel". In order to perform the tests more efficiently, many techniques such as the Taguchi method, design of experiments (DOE), neural and fuzzy theories are adopted to the proposed tests.