近年來,由於全球化競爭激烈與消費者的需求快速改變,產品樣式變化快速且多樣化,導致現今產品生命週期縮短。在產品開發過程需要大量的專業技術加上智慧財產權受到企業體的關注程度正逐年提升,因此專利迴避在產品開發過程至關重要。在過去需要耗費大量的人力和資源,進行專利搜尋並且逐篇閱讀內容,評估如何進行產品迴避設計。故本研究提出整合專利多元尺度分析與萃思方法論之產品創新設計流程之概念,針對重點專利並利用國際專利分類碼(IPC、UPC及FI)進行初步分類,並進行多元尺度分析,找出與目標專利相似度高之專利。並利用文字探勘技術結合倒傳遞類神經網路及使用者問題,透過TRIZ手法提供使用者產品迴避設計之方向。最後,本研究將整合方法流程進行系統建構,並將方法論具體應用在LED專利個案之創新設計上,以供參考。
In recent years, because the competition between enterprises has been more intense, and the preference of consumers has changed dramatically. The production changing rapidly and diversely of styles cut down the life cycle of products today. The enterprises that concern about the property right and lot of professional techniques in production processes are increasing year by year, so it's important to avoid patents issues in production developing processes. It took lot of manpower and resource to search, read patents and evaluate how to advance products patents avoiding designs in the past. We developed the concepts of production design processes for integrating the patent multi-scale analysis and TRIZ methodology in this study. First of all, we sorted out the key patents by the international patent classification codes, and then we found those patents similar with our target patents by the Multi-scale analysis. With the above described integrated approach, we used text mining technique and combined back-propagation neural network with user's problems to provide the design direction for users to avoid patent issues by TRIZ methods. Finally, this study will be integrated for building the system process method and the specific application of innovation in the LED design patent.