【正文】 基于CBR的排氣消聲器快速設(shè)計及評價方研究Research on CBRBased Rapid Design andEvaluation Method of Exhaust MufflerA Thesis Submitted to Chongqing Universityin Partial Fulfillment of the Requirement for theDegree of Master of EngineeringByXiang FeiSupervised by Prof. Deng ZhaoxiangMajor: Vehicle EngineeringCollege of Mechanical Engineering Chongqing University,Chongqing, ChinaMay2011I摘 要汽車排氣消聲器是多學科知識的有機結(jié)合體，各評價指標之間并不獨立，文中建立了排氣消聲器綜合性能評價體系，并結(jié)合領(lǐng)域?qū)＜抑R，實現(xiàn)了多指標的綜合定量評價，避免了采用單一指標進行逐一評價過程中產(chǎn)生的顧此失彼；排氣消聲器設(shè)計是弱理論強經(jīng)驗的設(shè)計過程，一個成熟的實例融合了大量的專家經(jīng)驗知識，基于實例的選型設(shè)計能夠快速的匹配出與設(shè)計目標最相近的實例，不但取代了消聲器原始設(shè)計方法中的概念設(shè)計，縮短了開發(fā)時間，還最大限度保留了專家設(shè)計經(jīng)驗知識，彌補了初級設(shè)計人員的經(jīng)驗不足。對排氣消聲器的各項性能要求所涉及的評價指標進行了詳細闡述，從排氣消聲器的傳遞損失自身性能、匹配性能、插入損失性能和壓力損失性能幾個方面對排氣消聲器的性能定量化評價方法進行了詳細的分析，建立了一種可量化的排氣消聲器綜合性能評價體系。對傳遞損失曲線進行頻段劃分和消聲量強弱等級劃分，從平均消聲量大小、消聲低谷和強消聲頻段等級的角度實現(xiàn)了傳遞損失的自身性能評價；通過對目標發(fā)動機排氣噪聲和各轉(zhuǎn)速下尾管噪聲目標曲線的基倍頻階次分析，得到了所需設(shè)計的排氣消聲器具有頻譜特性的消聲量目標曲線，將傳遞損失曲線與消聲量目標曲線進行比較，實現(xiàn)了傳遞損失的匹配性能評價；從插入損失值和各轉(zhuǎn)速下尾管噪聲達標情況兩個方面對排氣消聲器的插入損失性能進行了評價；根據(jù)額定工況下的壓力損失值和單位流量下的壓力損失值對不同結(jié)構(gòu)、不同系統(tǒng)的排氣消聲器空氣動力學性能進行了評價；在傳遞損失曲線平均消聲量和額定工況下插入損失值的評價過程中，綜合考慮了容積、質(zhì)量和連接管截面積的影響，引入了對排氣消聲器的工藝成本要求。采用定量與定性相結(jié)合的 AHP 決策方法獲取了排氣消聲器綜合性能評價指標體系的各項權(quán)重，完善了綜合評價體系，并通過實例驗證了評價體系的可操作性和可行性，為排氣消聲器的綜合性能評價和匹配設(shè)計提供了依據(jù)。對排氣消聲器設(shè)計過程及基于實例的推理（CBR）過程的特點進行了對比分析，發(fā)現(xiàn) CBR 推理過程與人類處理新問題的思維方式十分相近，論證了將 CBR應用于排氣消聲器設(shè)計的可行性；根據(jù) CBR 的推理過程提出了基于 CBR 的排氣消聲器快速設(shè)計思路和流程；詳細闡述了基于框架的排氣消聲器實例知識表達方法，在現(xiàn)有的排氣消聲器實例的基礎(chǔ)上建立了實例數(shù)據(jù)庫；結(jié)合排氣消聲器綜合性能評價體系，采用基于模糊相似優(yōu)先比的最大相似度實例檢索方法，從外形尺寸和性能要求方面實現(xiàn)了排氣消聲器的最優(yōu)實例的檢索，提出了匹配過程中實例數(shù)據(jù)的完整度概念，規(guī)定最優(yōu)實例的選擇應綜合考慮數(shù)據(jù)完整度、數(shù)據(jù)相似度和II優(yōu)先順序；采用基于規(guī)則推理和人機交互相結(jié)合的方法實現(xiàn)了對實例的修改；采用增量式實例保存策略實現(xiàn)了對實例的保存；采用人機交互的方法來實現(xiàn)實例數(shù)據(jù)庫的維護。最后通過 Visual 和 MATLAB 語言的聯(lián)合編程實現(xiàn)了基于 CBR 的排氣消聲器快速選型設(shè)計系統(tǒng)，利用所開發(fā)的系統(tǒng)完成了一款排氣消聲器的選型設(shè)計，初步驗證了系統(tǒng)的可操作性和實用性。關(guān)鍵詞：排氣消聲器，綜合評價，實例推理，快速設(shè)計IIIABSTRACTAutomobile exhaust muffler is the multidisciplinary bination withindispensible evaluation indexes. Based on expert knowledge, a prehensiveperformance evaluation system of exhaust muffler has been established, theprehensive quantitative evaluation of multiindex of exhaust muffler has beenimplemented, and the doubtfulness of paring the cases respectively according todifferent indexes has been avoided. Exhaust muffler design is a typical design processwith experience overweighs theory。 a mature case included large quantities of expertexperience knowledge. Based on mature cases, the most similar case can be identifiedvery quickly with casebased reasoning(CBR), which not only replaced conceptualdesign of the traditional method, but also retained expert experience knowledge as muchas possible, it shortened the development time, and maked up for the lack of experienceof Junior designer.The evaluation indexes which associate with all of the performance requirementsof exhaust muffler has been described in detail in this paper, and the quantitativeevaluation method of exhaust muffler also has been analysised in the aspects oftransmission loss, insert loss and pressure loss, then, a quantifiable prehensiveperformance evaluation system of exhaust muffler has been divided thetransmission loss curves by frequency band and by attenuation ability and achieved theselfevaluation of transmission loss front the points of average sound attenuation, lowtransmission loss, intensity transmission loss and so on. The sound attenuation targetcurve has been obtained through the order frequency analysis on the exhaust noise andthe goal curve of the tail pipe noise in each rotating speed, the transmission lossmatching evaluation was carried out by paring the transmission loss curve withsound attenuation target curve. The insert loss evaluation has been relized from thestudy of insert loss value and the reaching standard ability of the tail pipe noise in eachrotating speed. The aerodynamics evaluation indicators of exhaust muffler included thepressure loss of rated condition and unit mass evaluation system considered theinfluence of volume, quality and connecting pipe crosssectional area and also therequirement of process cost. The weights of evaluation system obtained by AHP, finally,the operability and feasibility of evaluation system has been verified through anexample, thus to provide foundation for exhaust muffler prehensive performanceIVevaluation and matching design.This paper demonstrated that applying CBR for muffler design is feasible byparing and analyzing the characteristics between exhaust muffler design processand casebased reasoning (CBR) process, proposed based on CBR exhaust muffler rapiddesign ideas and processes according to the reasoning process of CBR, expounded themuffler experience cases39。 knowledge representation method based on the framework,and established the instance database by existing cases，the optimal case search by thequalification of dimensions and performance has been carried out by the fuzzy analogypreferred ratio and the prehensive performance evaluation system. In this paper, theconcept of instance data integrity has been putted forward, and regulated that dataintegrity, data similarity and priorities should be taken into account when making theoptimal choice. The case Revise has been implemented based on Rulebased reasoning(RBR) and Human Computer Interaction (HIC)。 Incremental case saving is employed tosave cases and HIC to maintain the database.Finally, based on CBR rapid design system of exhaust muffler has beenimplemented by Visual and joint programming with MATLAB, pletedthe selection of exhaust muffler design with the assistance of the system, themaneuverability and practicability of selection system has been validated.Keywords：Exhaust muffler, Comprehensive evaluation, Casebased reasoning, Rapid目 錄中文摘要..........................................................................................................................................I英文摘要...............................................................................................................