能源消耗、污染排放與車輛使用之整合關聯模式研究(2/3)

Abstract

本計畫旨在建構能源消耗、污染排放與車輛使用之整合關聯模式，俾進一步評估及預測各種汽機車管理策略對與污染減量之效果。為達此一目的，本計畫擬利用資料探勘技術，針對公路監理單位攔、定檢資料庫加以分析，找出影響污染排放之重要解釋變數及污染排放之車型分群類別，並尋求相關污染排放參數之設定。本計畫將依選出之解釋變數及分群類別進行問卷設計，透過重覆施行之大規模問卷調查，以適當方法（如羅吉特模式、存活理論及馬可夫鏈）建立汽機車持有、使用及交易之家戶別個體模式，俾推估不同管理策略對汽機車持有、使用及交易之影響。復以聯立羅吉斯迴歸方法建立全國及地區別汽機車持有與使用之總體模式，俾推估車輛使用對能源消耗與污染排放之影響。最後，再利用雙層數學規劃方法，整合總體及個體模式，進行最佳化管理策略之模化、求解與驗證。為使決策者便於使用本整合關聯模式分析各種車輛管理策略對能源與污染減量之效果，本計畫擬進一步建立決策支援系統，並設計合適之使用者界面。

In this study, an integrated model that correlates vehicle usage with energy consumption and pollutant emissions will be developed to evaluate and forecast the effects of vehicle ownership and/or usage control strategies on the reduction of energy consumption and pollutant emissions. To achieve this goal, the study will employ data mining techniques, mainly based on the official database of aperiodic and periodic motor vehicle inspections, to extract key explanatory variables affecting emissions and to cluster similar vehicle emission types so as to properly set the pollutant parameters. According to the chosen explanatory variables and clustered vehicle types, proper questionnaires will be designed and a large-scale repeated household survey will be conducted to collect panel data with individual’s preferences, behaviors and potential responses of car and motorcycle ownership, transaction, and usage to the vehicle control strategies. This data will then be used to calibrate dynamic car and motorcycle ownership, transaction, and usage disaggregate models, using appropriate methods such as Logit modeling and Markov Chain. Aggregate models of car and motorcycle ownership and usage at national and regional levels will also be developed, using simultaneous logistic regression method, to correlate energy consumption and pollutant emissions with vehicle usage. In order to seek for optimal vehicle control strategies, a bi-level mathematic programming that incorporates all of these aggregate and disaggregate models will be formulated, solved and validated. At last, to facilitate the decision makers to examine the effects of energy and emissions reduction via the proposed integrated model, a decision support system with user-friendly interface will be developed.