人工神經(jīng)網(wǎng)絡(luò)外文翻譯

1、<p>　　附錄二：外文翻譯原文</p><p>　　The Kahnawake Survival School (KSS), located in the Kahnawake Mohawk Territory on the south shore of the St. Lawrence River across from Montreal, is more than a high school. It

2、 is a teaching tool for students, a community gathering place and a shelter in case of disaster. The 5500 m2 (60,000 ft2) building is composed of a central block with two wings, spread partially on two floors. The school

3、 opened in August 2008.</p><p>　　After a thorough inspection of the land owned by the Kahnawake Education Center (KEC), the design team chose a location in an existing clearing where the building would be ad

4、jacent to a forest but still accessible from the main highway of the reserve. The proposed location was analyzed by all building professionals to optimize the orientation/layout of the building. </p><p>　　Co

5、nsidering the age and history of the site, the decision was made to design a building and make it fit the site, preventing major site modifications and unnecessary cutting down of trees. This consideration was even more

6、important since an endangered tree species (Butternut) was present. Also, all soil removed from the site for excavation was piled near the highway to create a natural wall of dirt, grass and trees, to reduce the noise fr

7、om the highway.</p><p>　　Although building the facility in a clearing was a major construction challenge, it had many advantages for mechanical and electrical engineering. </p><p>　　About the Au

8、thor</p><p>　　Nicolas Lemire, P.Eng., is a principal at Pageau Morel and Associates, Montreal, QC, Canada.</p><p>　　Table 1: Average energy consumption for teaching facilities in 2006 (per Natur

9、al Resources Canada, OEE).</p><p>　　Table 2: Energy comparison of schools in Quebec, Ontario and Canada</p><p>　　Table 3: Construction costs in Canadian dollars.</p><p>　　For exampl

10、e, trees surrounding the path provide shade for the south fade of the facility during summer. With leaves shed in the winter, the resulting solar gains are used as an auxiliary heating source. </p><p>　　Tabl

11、e 1 shows data published by Natural Resources Canada on the average energy consumption of teaching facilities in 2006. </p><p>　　Table 2 shows an energy comparison between the average facilities presented in

12、 Table 1, the reference building as per the Model National Energy Code for Buildings (MNECB) of Canada for a teaching facility including geothermal energy in the Province of Quebec1 and in the Province of Ontario,2 and t

13、he new Kahnawake Survival School. </p><p>　　Table 2 shows that KSS performs well at 66.8% more efficient than the average teaching facility in the Province of Quebec and 51.4% more efficient than the referen

14、ce building of the MNECB of Canada for a teaching facility in Quebec using geothermal energy.</p><p>　　Heat recovery is applied to fresh air and exhaust using an enthalpy wheel. Coils and filters in the syst

15、ems are selected at 1.8 m/s (350 fpm) to reduce static pressure loss. Variable frequency drives are installed on all fans and motors and are directly coupled to avoid belt losses and reduce maintenance. Motors are high e

16、fficiency. A closed loop geothermal heat exchanger (15 vertical boreholes of 137 m2 [450 ft]) is used to supply tempered glycol to local geothermal water-to-air heat pumps. A w</p><p>　　Additional energy eff

17、icient measures implemented are high efficiency lighting fixtures and fresh air control using CO2 sensors for the gym.</p><p>　　Ducting was designed to group the classrooms in four zones, with a fifth zone f

18、or the offices. As soon as all classrooms in one zone are unoccupied, that ventilation zone is shut down. When all zones are closed, the main system is turned off.</p><p>　　Physical separations are present b

19、etween classrooms and offices, giving a means of controlling IAQ and energy consumption depending on occupation modes (offices are used in summer, while classes are not; classes can be used at night throughout the year w

20、hile offices are empty at night, etc.).</p><p>　　Commissioning was performed with emphasis on performances of primary air-handling units (AHUs) and ventilation strategies. To deal with the amount of fresh ai

21、r injected into the gymnasium, CO2 sensors were installed in the return duct leading to the air-handling units, analyzing CO2 quantities contained in return air. Fresh air is injected in the mixing box of the AHU to main

22、tain CO2 levels at 800 ppm. </p><p>　　The special variable air volume (VAV) system with predetermined outside air rate and terminal reheat is an efficient way of providing effective indoor environmental qual

23、ity to the users. All zones in the building can maintain effective temperature within the ASHRAE comfort zone as defined in ASHRAE Standard 55. A minimum humidity level of 30% is maintained during winter using electrical

24、 steam generator humidifiers installed in the air-Image: . 2009 DigitalGlobe</p><p>　　Image: . 2009 DigitalGlobe technology award case studies</p><p>　　Building at a Glance</p><p>　

25、　Name: Kahnawake Survival School</p><p>　　Location: Kahnawake, QC, Canada</p><p>　　Owner: Kahnawake Education Center</p><p>　　Principal Use: School</p><p>　　Includes: H

26、igh School, Community </p><p>　　Center, Public Assembly</p><p>　　Employees/Occupants: 450 students</p><p>　　Gross Square Footage: 60,000 ft2</p><p>　　Substantial Comple

27、tion/Occupancy: August 2008</p><p>　　Occupancy: 100%handling units. During summer, a maximum of 60% is allowed (design criteria for offices).</p><p>　　Considering the many types of activity occu

28、rring in the building (teaching, administration, community activities, shows, sporting events, community meetings and shelter), two basic options were analyzed: dedicated systems and centralized systems. After analysis,

29、it was decided to combine both strategies. The use of a centralized system to condition the amount of outside air required and send it in all zones was the best solution. The capability to operate at variable flow was a

30、major aspect of t</p><p>　　Because classrooms are not used during the hotter months of the year (from mid-June to the end of August), cooling the classrooms and the gym was questioned. It was decided to use

31、local geothermal water-to-air heat pumps into the administrative and office areas and in the cafeteria/student lounge as those areas were more likely to be used throughout the year or during summer for events. The fresh

32、air system was equipped with a geothermal water-to-water heat pump to allow heating/cooling/dehumidi</p><p>　　classrooms and the gym. </p><p>　　For the gym, a provision has been made to allow in

33、stallation of cooling capacity in the system in the future by adding a water-water heat pump to supply a coil. Natural ventilation is available for all classrooms and the gym using operable windows. The main central corr

34、idor is open on two stories and continues higher (almost three floors) to act as a natural chimney. All classrooms are opened to the central corridor (using operable panels). When the outside conditions are adequate, a &

35、lt;/p><p>　　special green light shows teachers/users it is a good day to use natural ventilation. Operable windows located at the top of the natural chimney are opened, and teachers/users can decide if they wan

36、t to open them.</p><p>　　If very hot days occur during the academic year, a provision for two propeller fans, located at each end of the main corridor, was planned (at the top of the natural chimney). This w

37、ould force air movement through the building (using natural ventilation openings in classrooms but closing the windows at the top of the natural chimney). The same strategy was applied to the gym, allowing it to be natur

38、ally cooled. Also, a dedicated air-handling unit was installed into the gym for ventilation and he</p><p>　　A centralized building automation system (BAS) links all mechanical components through a centralize

39、d DDC network. A central panel is located in the main mechanical room and is simple to use so that occupants who are present outside of normal business hours can start/stop different features of the building (natural ven

40、tilation, forced natural ventilation fans, primary fresh air system, gym ventilation system and gym forced natural ventilation fans).</p><p>　　Commissioning was done on the BAS, which helped improve energy e

41、fficiency. This process continued after delivery of thebuilding and will continue for a few years to perfectly tune the building to the desired operation.</p><p>　　Capital costs were controlled by providing

42、simple systems that rely on well-established, low-cost technologies and by optimizing equipment selection for dependability, low maintenance and maximum efficiency. A major advantage of the VAV systems with terminal rehe

43、at is that, despite different load requirements, a comfortable environment can be maintained in all rooms. This makes the systems flexible enough to adapt, at low cost, to any layout modification.</p><p>　　D

44、esigning complicated systems is not always a guarantee for energy efficiency. In fact, the guiding principle is that simpler systems (as long as energy efficiency is not compromised) are understood better by maintenance

45、personnel, which lowers operation costs.</p><p>　　The decrease of energy consumption led to a reduction of CO2 emissions by about 192 Mg (212 tons) per year. The total energy consumption reduction per year c

46、orresponds approximately to the energy consumption of 92 average houses.</p><p><b>　　外文翻譯中文</b></p><p>　　本文研究人工神經(jīng)網(wǎng)絡(luò)（ANN）的適用性，如汽車(chē)空調(diào)系統(tǒng)（AAC）的適用性能預(yù)測(cè)。制冷劑使用的是HFC134a。為了這個(gè)目的，實(shí)驗(yàn)廠原始組件從乘用車(chē)空調(diào)系統(tǒng)

47、的一個(gè)標(biāo)準(zhǔn)的尺寸制造。實(shí)驗(yàn)系統(tǒng)是在穩(wěn)定條件下操作，同時(shí)改變壓縮機(jī)的轉(zhuǎn)速，冷卻能力和冷凝溫度。然后，使用了一些實(shí)驗(yàn)數(shù)據(jù)進(jìn)行模擬，系統(tǒng)的人工神經(jīng)網(wǎng)絡(luò)模型，基于標(biāo)準(zhǔn)的反向傳播算法的開(kāi)發(fā)。該模型被用于各種性能參數(shù)的預(yù)測(cè)系統(tǒng)，即壓縮機(jī)的功率，在冷凝器的散熱率，制冷劑的質(zhì)量流量，壓縮機(jī)排出溫度和性能系數(shù)。這些參數(shù)的人工神經(jīng)網(wǎng)絡(luò)預(yù)測(cè)通常約定好在0.968-0.999的范圍內(nèi)，相關(guān)系數(shù)的實(shí)驗(yàn)值，在范圍內(nèi)的平均相對(duì)誤差1.52-2.5％非常低的根均方誤差

48、。這項(xiàng)研究表明，空調(diào)系統(tǒng)，即使是那些采用可變速度壓縮機(jī)，如AAC系統(tǒng)，也可以使用具有高度準(zhǔn)確性的人工神經(jīng)網(wǎng)絡(luò)建模準(zhǔn)確度。</p><p>　　2005年愛(ài)思唯爾有限公司保留所有權(quán)</p><p><b>　　1 介紹</b></p><p>　　汽車(chē)空調(diào)系統(tǒng)（AAC），通常采用蒸氣壓縮式制冷電路，目前使用的HFC134a作為制冷劑，在乘客車(chē)廂內(nèi)

49、實(shí)現(xiàn)夏季舒適。AAC系統(tǒng)由于壓縮機(jī)是由發(fā)動(dòng)機(jī)驅(qū)動(dòng)的皮帶帶動(dòng)，壓縮機(jī)的轉(zhuǎn)速與發(fā)動(dòng)機(jī)的轉(zhuǎn)速成正比，這將影響該系統(tǒng)的制冷能力，以改變發(fā)動(dòng)機(jī)轉(zhuǎn)速的函數(shù)。因此，這些系統(tǒng)不同于國(guó)內(nèi)的空調(diào)系統(tǒng)，由于不同的壓縮機(jī)轉(zhuǎn)速和散熱能力以及不穩(wěn)定的制冷負(fù)荷，這些復(fù)雜特性影響AAC系統(tǒng)建模，因此采用經(jīng)典的技術(shù)。</p><p>　　AAC系統(tǒng)公開(kāi)發(fā)表的文獻(xiàn)是非常有限的，因?yàn)檫@個(gè)行業(yè)是一個(gè)競(jìng)爭(zhēng)激烈的行業(yè)。 專(zhuān)家 發(fā)表過(guò)的文章。 Jung

50、[1]一些混合制冷劑電腦分析，如HFC134a，HCFC142b，RE170，HC290和HC600a盡可能替代CFC12在現(xiàn)有的AAC系統(tǒng)。他們的分析結(jié)果初步篩選替代品。然后，他們根據(jù)實(shí)驗(yàn)的結(jié)果提出的替代制冷劑的混合物。通過(guò)計(jì)算機(jī)分析，發(fā)現(xiàn)一個(gè)HFC134a/RE170的混合物是最好的混合物替代CFC12。但是，他并沒(méi)有報(bào)告理論和實(shí)際之間的任何實(shí)驗(yàn)比較結(jié)果。</p><p>　　李和Yoo [2] 對(duì)制冷劑為

51、HFC134a的AAC系統(tǒng)的各組成部分進(jìn)行了分析，并開(kāi)發(fā)了一個(gè)仿真模型，開(kāi)發(fā)出整個(gè)系統(tǒng)相結(jié)合的組件的單獨(dú)性能分析程序。他們的計(jì)劃用于研究蒸發(fā)器的性能是基于實(shí)驗(yàn)結(jié)果，該程序用于研究冷凝器性能是假設(shè)沒(méi)有過(guò)冷冷凝器出口。他們發(fā)現(xiàn)整個(gè)系統(tǒng)的仿真模型和實(shí)驗(yàn)結(jié)果之間的誤差是在7％以?xún)?nèi)。</p><p>　　ratts和Brown [3]實(shí)驗(yàn)對(duì)性能的影響進(jìn)行了分析，制冷劑為四氟乙烷（HFC134a）裝料水平的AAC的系統(tǒng)。為

52、了這個(gè)目標(biāo)，他們確定了單個(gè)組件在一個(gè)AAC系統(tǒng)的損失作為一個(gè)功能的制冷劑，對(duì)他的的電荷使用第二定律。他們發(fā)現(xiàn)，壓縮機(jī)和冷凝器部件造成的總損失的比例最大，而蒸發(fā)器和擴(kuò)展設(shè)備的損失占一個(gè)較小比例的損失。</p><p>　　Rabghi和尼牙孜 [4]加裝一個(gè)的CFC12 的AAC系統(tǒng)使用的制冷劑為HFC134a，并確定了實(shí)驗(yàn)壓縮機(jī)速度的函數(shù)的系統(tǒng)參數(shù)，在各制冷劑的性能系數(shù)（COP）下，他們發(fā)現(xiàn)使用CFC12 AAC

53、系統(tǒng)比使用四氟乙烷（HFC134a）的系統(tǒng)有一個(gè)更好的COP。jabardo等。 [5] 開(kāi)發(fā)了一種含有一個(gè)可變的AAC系統(tǒng)的穩(wěn)態(tài)仿真模型容量壓縮機(jī)，微通道平行流冷凝器，恒溫膨脹閥和板翅片管式蒸發(fā)器。他們測(cè)試了一個(gè)實(shí)驗(yàn)單元上的模型的有效性。他們發(fā)現(xiàn)，仿真和實(shí)驗(yàn)結(jié)果之間的偏差的制冷量，COP和制冷劑質(zhì)量為壓縮機(jī)速度的函數(shù)的流量分別為5％以?xún)?nèi)。然而，相同的性能參數(shù)，作為相對(duì)于蒸發(fā)器的回風(fēng)溫度對(duì)模擬結(jié)果的偏差的函數(shù)實(shí)驗(yàn)高達(dá)18％。</p

54、><p>　　joudi等。 [6]模擬了一個(gè)理想的AAC系統(tǒng)的工作性能與幾個(gè)制冷劑確定最合適的替代制冷劑為CFC12。他們的模型預(yù)測(cè)的混合物HC290/HC600a的最佳替代CFC12。之后，他們比較了各種性能使用CFC12的實(shí)驗(yàn)AAC系統(tǒng)的參數(shù)和的混合物HC290/HC600a為工作液體。他們觀察到，壓縮機(jī)的功率消耗在HC290/HC600a的情況下小幅走高比中的CFC12的情況下為相同的冷卻能力。然而，他們沒(méi)有

55、報(bào)告任何比較仿真和實(shí)驗(yàn)結(jié)果。</p><p>　　Kaynakli Horuz [7]分析了HFC134a的AAC系統(tǒng)性能的實(shí)驗(yàn)，以找到最佳的工作條件。他們提出了一些性能參數(shù)，如冷卻能力，壓縮機(jī)功率，總功率消耗，制冷劑的質(zhì)量流率和COP作為冷凝溫度，蒸發(fā)器的回流空氣溫度的函數(shù)，環(huán)境溫度和壓縮機(jī)轉(zhuǎn)速。最近的研究旨在降低全球氣候變暖源于AAC系統(tǒng)設(shè)計(jì)或者修改的系統(tǒng)需要較少量的HFC134a的或新穎的系統(tǒng)使用不同的制冷

56、劑如CO2和碳?xì)浠衔铩０偷偬幚頋撛诘腍FC134a增強(qiáng)的AAC系統(tǒng)，降低全球變暖的影響</p><p>　　[8]為此，他調(diào)查增加了壓縮機(jī)的等熵效率的影響，增加冷凝器的有效性，在蒸發(fā)器中的空氣側(cè)壓降降低，增加的冷凝器空氣流和降低系統(tǒng)的COP的制冷負(fù)載。從基線四氟乙烷（HFC134a）系統(tǒng)獲得的實(shí)驗(yàn)結(jié)果，從切實(shí)增強(qiáng)四氟乙烷（HFC134a）系統(tǒng)的比較的基礎(chǔ)上，他指出的增強(qiáng)的系統(tǒng)可能是最實(shí)際的方法來(lái)處理AAC 系統(tǒng)

57、等引起的全球變暖。</p><p>　　[9]研究了CO2和HFC134a AAC系統(tǒng)使用半理論周期模型的性能優(yōu)劣。在除了標(biāo)準(zhǔn)制冷回路組件，即壓縮機(jī)，冷凝器，膨脹裝置和蒸發(fā)器，其二氧化碳的系統(tǒng)配備的液體線/吸氣管道熱交換器。</p><p>　　[10]他們確定四氟乙烷（HFC134a）具有更好的COP比二氧化碳的COP差距是依賴(lài)于壓縮機(jī)的轉(zhuǎn)速和環(huán)境溫度為了找到一個(gè)合適的烴的替代，具有較低

58、的全球變暖潛力比四氟乙烷（HFC134a），Ghodbane模擬性能AAC使用碳?xì)渲评鋭碒C152a，HC270，HC290 HC600a的，在COP和壓縮機(jī)排氣溫度的系統(tǒng)。他們決定將HC152a和HC270制冷劑的HFC134a系統(tǒng)，分別為11％和15％，而HC290顯示，只有輕微改善。然而，由于其潛在的易燃性，AAC系統(tǒng)使用碳?xì)渲评鋭┰诒徽J(rèn)為是不安全的，除非一些額外的設(shè)計(jì)注意事項(xiàng)。</p><p>　　t

59、aken.Tian和Li [11] 開(kāi)發(fā)了一個(gè)數(shù)學(xué)模型的HFC134a AAC系統(tǒng)與可變?nèi)萘繅嚎s機(jī)，模擬其穩(wěn)定狀態(tài)下的性能。他們的模型中確定的壓縮機(jī)速度，環(huán)境溫度和蒸發(fā)壓力上的蒸發(fā)器的空氣流率的影響，冷凝壓力，冷卻能力，并表示壓縮機(jī)的功率。他們證實(shí)了一個(gè)實(shí)驗(yàn)單元上的模型的結(jié)果，發(fā)現(xiàn)，在11％之內(nèi)的模擬和測(cè)量的參數(shù)之間的偏差。</p><p>　　Hosoz和Direk [12]處理的操作與該特征的四氟乙烷（HFC

60、134a）AAC系統(tǒng)的性能特征為空氣，以空氣能熱泵。為了這個(gè)目的，他們開(kāi)發(fā)了一個(gè)實(shí)驗(yàn)系統(tǒng)，并測(cè)試了它在空調(diào)和熱泵模式，改變壓縮機(jī)的轉(zhuǎn)速和入口空氣溫度的室外和室內(nèi)的線圈。他們?cè)u(píng)估了性能的集成系統(tǒng)中的制冷和制熱能力，COP，壓縮機(jī)的排氣溫度和速度破壞該系統(tǒng)的每個(gè)組件。他們決定，通常熱泵運(yùn)行產(chǎn)生了較高的COP和更低的價(jià)格相比，每單位容量（火用）破壞的空調(diào)操作，雖然不足heating.From，這個(gè)簡(jiǎn)短的文獻(xiàn)綜述，可以觀察到，一些研究者傾向于開(kāi)

61、發(fā)數(shù)學(xué)模型，以確定AAC系統(tǒng)的各種性能參數(shù)，而其他人為了同樣的目的進(jìn)行徹底和昂貴的實(shí)驗(yàn)研究。在傳統(tǒng)的建模方法，所采用的計(jì)算機(jī)模擬是復(fù)雜和耗時(shí)的，由于其處理復(fù)雜的微分方程的解。此外，該數(shù)學(xué)模型，需要大量的幾何參數(shù)，定義系統(tǒng)，這可能不是可用的，并在許多情況下，他們的預(yù)測(cè)可能不夠準(zhǔn)確。這兩種方法的替代，AAC系統(tǒng)可以模擬人工神經(jīng)網(wǎng)絡(luò)（ANN），大大減少工程工作。這種新的建模技術(shù)是基于模仿人類(lèi)大腦的結(jié)構(gòu)和機(jī)制，被用在越來(lái)越多的工程應(yīng)用中的經(jīng)典方

62、法失敗或過(guò)于復(fù)雜，是used.ANNs允許的造型在復(fù)</p><p>　　利用從實(shí)驗(yàn)獲得的穩(wěn)定的狀態(tài)數(shù)據(jù)AAC系統(tǒng)，一個(gè)系統(tǒng)的神經(jīng)網(wǎng)絡(luò)模型已經(jīng)研制成功。與使用該模型，各種性能參數(shù)的系統(tǒng)，即壓縮機(jī)的功率，在冷凝器的散熱率，制冷劑的質(zhì)量流量，壓縮機(jī)排出溫度和COP，已預(yù)測(cè)和與實(shí)際的相比。</p><p><b>　　2 人工神經(jīng)網(wǎng)絡(luò)</b></p><

63、p>　　人工神經(jīng)網(wǎng)絡(luò)在電腦的方式訴諸人的行為為基礎(chǔ)的學(xué)習(xí)機(jī)制，以反映大腦的功能。利用 樣品從實(shí)驗(yàn)中，神經(jīng)網(wǎng)絡(luò)可以應(yīng)用于不帶算法的??解決方案，或與太復(fù)雜的算法以找到的解決方案的問(wèn)題。他們的學(xué)習(xí)能力的例子人工神經(jīng)網(wǎng)絡(luò)更靈活，更強(qiáng)大的比參數(shù)的方法[25]。 </p><p>　　人工神經(jīng)網(wǎng)絡(luò)是由大量被稱(chēng)為神經(jīng)元相互連接的處理節(jié)點(diǎn)。每個(gè)神經(jīng)元接受輸入和第一形式的加權(quán)組的

64、總和以偏置的加權(quán)輸入定義的[26]</p><p><b>　　(1)</b></p><p><b>　　(2)</b></p><p>　　其中P和的元素?cái)?shù)目和輸入向量，分別和b的互連權(quán)重是對(duì)神經(jīng)元的偏置。請(qǐng)注意，作為偏壓的互連權(quán)重的一組神經(jīng)元的知識(shí)被存儲(chǔ)在。然后，神經(jīng)元的輸出響應(yīng)。為此，總和的加權(quán)輸入以偏置處理通過(guò)激

65、活函數(shù)，由f表示，和計(jì)算的輸出基本上是，神經(jīng)元模型模擬生物神經(jīng)元，將觸發(fā)時(shí)，顯著激發(fā)，即神經(jīng)元的輸入，n，是足夠大的。有許多方法來(lái)定義，如閾值函數(shù)的激活函數(shù)，S型函數(shù)和雙曲正切函數(shù)。</p><p>　　使用一個(gè)合適的學(xué)習(xí)方法中，人工神經(jīng)網(wǎng)絡(luò)訓(xùn)練的處理節(jié)點(diǎn)之間的連接，即加權(quán)系數(shù)的值，通過(guò)調(diào)整，以執(zhí)行特定的功能。訓(xùn)練過(guò)程繼續(xù)進(jìn)行，直到網(wǎng)絡(luò)的輸出相匹配的目標(biāo)，即所需的輸出。網(wǎng)絡(luò)之間的誤差輸出和所需的輸出最小化，通過(guò)修

66、改的權(quán)重和偏見(jiàn)。當(dāng)誤差低于一個(gè)確定的值或超過(guò)歷元的最大數(shù)目，在訓(xùn)練過(guò)程被終止。然后，將該訓(xùn)練的網(wǎng)絡(luò)可以使用模擬系統(tǒng)的輸出的輸入，但沒(méi)有之前引入。</p><p>　　通常分為三個(gè)部分：一個(gè)輸入層，隱層和輸出層的人工神經(jīng)網(wǎng)絡(luò)的架構(gòu)。在輸入層中包含的信息被映射到通過(guò)隱藏layers.Each神經(jīng)元的輸出層，可以只將其輸出發(fā)送到上級(jí)層的神經(jīng)元上，并只從接收其輸入端的較低層的神經(jīng)元?；谌斯ど窠?jīng)網(wǎng)絡(luò)預(yù)測(cè)的性能評(píng)價(jià)的網(wǎng)絡(luò)輸

67、出之間的回歸分析，即預(yù)測(cè)的參數(shù)，和相應(yīng)的目標(biāo)，即實(shí)驗(yàn)值。該標(biāo)準(zhǔn)用于測(cè)量網(wǎng)絡(luò)性能的相關(guān)系數(shù)，平均相對(duì)誤差和均方根方誤差。相關(guān)系數(shù)評(píng)估預(yù)測(cè)和實(shí)驗(yàn)結(jié)果之間的關(guān)系的強(qiáng)度。這個(gè)系數(shù)之間的實(shí)際值與預(yù)測(cè)輸出[27]</p><p><b>　　(3)</b></p><p>　　其中，是a和p集指的實(shí)際輸出（實(shí)驗(yàn)）和預(yù)測(cè)的輸出集合，分別之間的協(xié)方差，和[27]被定義</p&g

68、t;<p><b>　　(4)</b></p><p>　　其中E是期望值，la是平均的值的一組和lp的p集是平均值。如果，則a和p表示以不相關(guān)。同樣，和是a和p的套，相應(yīng)的汽車(chē)的協(xié)方差，1和1之間的相關(guān)系數(shù)的范圍由下式給出。 </p><p><b>　　(5)</b></p><p><b>　

69、　(6)</b></p><p>　　R值接近1表示一個(gè)更強(qiáng)大的正的線性關(guān)系，而R值接近1表示一個(gè)較強(qiáng)的負(fù)相關(guān)關(guān)系。</p><p>　　的平均相對(duì)誤差，這是錯(cuò)誤和實(shí)驗(yàn)值之間的平均比值，計(jì)算從</p><p><b>　　(7)</b></p><p>　　其中N是數(shù)據(jù)集合中的點(diǎn)的數(shù)量。</p>

70、<p>　　最后，均方根誤差給出</p><p><b>　　(8)</b></p><p>　　3 實(shí)驗(yàn)裝置和測(cè)試程序的說(shuō)明</p><p>　　神經(jīng)網(wǎng)絡(luò)建模已被應(yīng)用到的實(shí)驗(yàn)AAC系統(tǒng)的蒸氣壓縮式制冷電路和作為制冷劑使用的HFC134a。實(shí)驗(yàn)系統(tǒng)中，如圖所示1，主要由與原來(lái)的組件從汽車(chē)標(biāo)準(zhǔn)尺寸空調(diào)系統(tǒng)。該系統(tǒng)的制冷回路包括一個(gè)五

71、氣缸旋轉(zhuǎn)斜盤(pán)壓縮機(jī)，微通道平行流冷凝器，內(nèi)部均衡的恒溫膨脹閥，層疊式蒸發(fā)器和一種液體的接收器/干燥器。</p><p>　　的蒸發(fā)器保持在其原始的塑料外殼，并插入一個(gè)空氣導(dǎo)管上游的一個(gè)模擬的乘客艙。此室具有大約1.5立方米的容積的，并包含電加熱器，以承載系統(tǒng)的熱負(fù)荷。該加熱器可以控制510和3060 W之間的間隔為510 W.的空氣流被分發(fā)由鼓風(fēng)機(jī)的原來(lái)的AAC系統(tǒng)的蒸發(fā)器，乘客室和回風(fēng)管組成的閉合電路。<

72、/p><p>　　壓縮機(jī)是皮帶驅(qū)動(dòng)由三相4千瓦的電動(dòng)機(jī)與標(biāo)稱(chēng)轉(zhuǎn)速為2850轉(zhuǎn)，經(jīng)由變頻器通電的電動(dòng)機(jī)，在各壓縮機(jī)的速度來(lái)操作該系統(tǒng)。一個(gè)AAC系統(tǒng)的冷卻能力，通常是由一個(gè)恒溫器，通電的電磁離合器脫開(kāi)壓縮機(jī)軸的旋轉(zhuǎn)滑輪，當(dāng)所需的回流空氣溫度來(lái)實(shí)現(xiàn)控制。然而，為了不中斷，以測(cè)試系統(tǒng)在穩(wěn)態(tài)操作中，實(shí)驗(yàn)系統(tǒng)沒(méi)有配備一個(gè)恒溫器。冷凝器風(fēng)扇電機(jī)是具有可變輸出電壓由直流電源通電。由于經(jīng)過(guò)冷凝器的空氣流率依賴(lài)于風(fēng)扇電機(jī)兩端的電壓，改

73、變?cè)撾妷涸试S獲得廣泛的冷凝溫度，而不管在冷凝器入口的空氣溫度。</p><p>　　原來(lái)的系統(tǒng)的制冷劑軟管由一種彈性材料和絕緣也中采用的實(shí)驗(yàn)系統(tǒng)。由聚氨酯泡沫體絕緣的空氣管道和乘客車(chē)廂具有厚度為5厘米，該系統(tǒng)被指控的HFC134a為700克。</p><p>　　進(jìn)行測(cè)量的類(lèi)型和位置，也表示在圖中1 K型熱電偶的制冷劑和空氣溫度檢測(cè)系統(tǒng)的關(guān)鍵點(diǎn)。鋁管在軟管的外表面直接接觸的制冷劑溫度的熱

74、電偶。無(wú)論是在入口和出口的蒸發(fā)器的空氣流的干球和濕球溫度進(jìn)行測(cè)定，在三個(gè)不同的位置，使蒸發(fā)器的出口處的空氣溫度進(jìn)行檢測(cè)，并且，找到它們的平均值，吸氣和排氣測(cè)壓力，這是假定的蒸發(fā)和冷凝壓力的吸氣和排氣壓力相等，分別掠過(guò)蒸發(fā)器的空氣的質(zhì)量流率是通過(guò)測(cè)量由風(fēng)速計(jì)回風(fēng)管中的平均空氣速度，確定找到的空氣密度的幫助下，在蒸發(fā)器入口的干，濕球溫度測(cè)量和評(píng)價(jià)它們?cè)谘刂c所述管道的流通面積的連續(xù)性方程。所述壓縮機(jī)的旋轉(zhuǎn)速度是由數(shù)字式轉(zhuǎn)速計(jì)測(cè)量通過(guò)測(cè)量加熱

75、器兩端的電壓和電流消耗，而被發(fā)現(xiàn)的熱量輸入到乘客車(chē)廂?？偨Y(jié)于表1中的某些功能的儀表。</p><p>　　在測(cè)試過(guò)程中，系統(tǒng)上的進(jìn)入冷凝器的空氣的溫度和速度的組合效應(yīng)的表現(xiàn)表示的冷凝溫度。在不同的環(huán)境溫度和冷凝器的空氣流速進(jìn)行該測(cè)試，得到兩個(gè)不同的冷凝溫度為50</p><p>　　60攝氏度。通過(guò)改變的熱量輸入到隔室為所需的值，調(diào)整所述壓縮機(jī)的速度，通過(guò)改變逆變器的輸出頻率，并通過(guò)不同的

76、輸出電壓的直流電流，冷凝器風(fēng)扇的速度的改變來(lái)實(shí)現(xiàn)所需的冷凝溫度動(dòng)力源。</p><p>　　該實(shí)驗(yàn)系統(tǒng)操作在穩(wěn)定狀態(tài)下，通過(guò)改變隔室1530和3060W之間的間隔與兩個(gè)冷凝溫度為510 W的熱負(fù)荷。對(duì)于每個(gè)熱負(fù)荷和冷凝溫度，壓縮機(jī)的轉(zhuǎn)速變化，在600和1400轉(zhuǎn)之間的間隔為200rpm。由于AAC壓縮機(jī)依靠飛濺潤(rùn)滑，最小的壓縮機(jī)轉(zhuǎn)速600轉(zhuǎn)，以避免被選為潤(rùn)滑不足可能出現(xiàn)在較低的速度。由于沒(méi)有配備一個(gè)恒溫器，實(shí)驗(yàn)系