資源描述:
《基于模糊認(rèn)知圖的熱網(wǎng)控制的研究》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫(kù)。
1、大連理工大學(xué)碩士學(xué)位論文摘要隨著集中供熱覆蓋面積的增加,研究集中供熱網(wǎng)控制算法的優(yōu)化問(wèn)題,提高供熱質(zhì)量,具有重要意義。當(dāng)前我國(guó)集中供熱面臨許多問(wèn)題,熱網(wǎng)控制技術(shù)較落后,資源浪費(fèi)嚴(yán)重。為了提高供熱效率,減少資源浪費(fèi),達(dá)到均勻供熱的目的,本文探討了熱網(wǎng)控制算法的優(yōu)化問(wèn)題,充分結(jié)合熱網(wǎng)實(shí)際控制操作中“粗調(diào)”和“微調(diào)"的特點(diǎn),以實(shí)際熱網(wǎng)工藝要求為背景,提出具有推理能力的控制策略。熱網(wǎng)是一種結(jié)構(gòu)復(fù)雜、規(guī)模龐大的城市管網(wǎng)系統(tǒng),具有大慣性、大滯后性、非線性和強(qiáng)耦合性等特點(diǎn),不易建立準(zhǔn)確的數(shù)學(xué)模型,常規(guī)控制方法很
2、難達(dá)到控制要求。本文介紹了熱網(wǎng)的基本概念理論,將新穎的FCM(fuzzycognitivemap,模糊認(rèn)知圖)算法應(yīng)用于熱網(wǎng)控制中,基于實(shí)驗(yàn)室設(shè)備構(gòu)建FCM模型,實(shí)現(xiàn)基于模型的FCM控制。FCM是神經(jīng)網(wǎng)絡(luò)和模糊控制相結(jié)合的產(chǎn)物,通過(guò)結(jié)構(gòu)表示知識(shí)。FCM通過(guò)NHL(nonlinearhebbianlearning,.非線性赫布學(xué)習(xí))進(jìn)行學(xué)習(xí)調(diào)整,但普通的NHL算法是無(wú)監(jiān)督的學(xué)習(xí)訓(xùn)練。本文根據(jù)熱網(wǎng)的特點(diǎn)改進(jìn)NHL,提出有監(jiān)督的NHL,實(shí)現(xiàn)對(duì)推理過(guò)程的可控調(diào)節(jié),并針對(duì)熱網(wǎng)控制的特點(diǎn),在構(gòu)建FCM時(shí),結(jié)合
3、NHL和專家知識(shí)兩種構(gòu)建方法,有效利用專家經(jīng)驗(yàn)建立初始FCM,再利用NHL實(shí)現(xiàn)FCM自動(dòng)調(diào)整,使FCM能準(zhǔn)確表達(dá)實(shí)際系統(tǒng)。本文基于FCM模型構(gòu)建實(shí)驗(yàn)室熱網(wǎng)控制平臺(tái),將FCM模型嵌入實(shí)驗(yàn)室熱網(wǎng)控制算法中,實(shí)現(xiàn)基于模型的控制策略。通過(guò)理論分析與實(shí)驗(yàn)驗(yàn)證,證明基于FCM模型的控制算法能夠滿足熱網(wǎng)控制要求,縮短熱網(wǎng)控制時(shí)間,并提高熱網(wǎng)系統(tǒng)的控制精度和穩(wěn)定性,使熱網(wǎng)系統(tǒng)狀態(tài)滿足工藝要求,實(shí)現(xiàn)均勻供熱。關(guān)鍵詞:模糊認(rèn)知圖;熱網(wǎng);均勻供熱;非線性赫布學(xué)習(xí)基于模糊認(rèn)知圖的熱網(wǎng)控制的研究Researchofheat
4、ingnetworkcontrolbasedontheFuzzyCognitiveMapsAbstractWiththeincreaseofthedistrictheatingsupplyarea,itisofgreatsignificancetoresearchonoptimizingthecontrolarithmeticofcentralheatingsupplynetwork,andimprovesheatingeffect.Nowadays,thedistrictheatingsuppl
5、yhaslotsofproblemsinourcountry,andwasteofmuchresource.Inordertobemoreefficient,andthetemperatureCanbecontrolledatanevenlevel.,thethemeofoptimizingthecontrolarithmetichasbeendiscussed,combiningthepointsof”coarsecontrol”and”finecontrol”duringtheprocess.
6、putsforwardtoacontrolmethodwhichhasillationabilitywiththetechniquesrequirementsofheatingsupplynetwork.Heatingsupplynetworkisacitypipenetworksystemwithcomplexstructureandlargescale.Ithasmanycharacteristics,likelargeinertia,largehysteresis,nonlinear,str
7、ongcouplingandSOon,soit’Sdifficulttoconstructthemodelexactly,andnormalcontrolmethodsaredifficulttosatisfytherequirements.ThispaperintroducesthebasictheoryandtheFCM(fuzzycognitivemap)arithmetic,constructsFCMmodeloftheinstrumentinlaboratorytoachievethec
8、ontrolgoalbyFCM.FCMcomesfromthecombinationofNeuralNetworksandfuzzycontrol,expressesinformationbystructure.FCMlearnsbyNHL(nonlinearhebbianlearning),andbutnormalNHLisunsupervisedlearning,SOsupervisedlearningNHLisfoundtoachievestheadjustmentofthe