【摘要】：在醫(yī)院建筑的用能計(jì)量中,大量的能耗數(shù)據(jù)被收集并存儲(chǔ),在其中往往隱含著許多有價(jià)值的信息。本課題在醫(yī)院用能統(tǒng)計(jì)工作已完成的前提下,從所收集的原始數(shù)據(jù)出發(fā),通過(guò)數(shù)據(jù)挖掘方法,針對(duì)醫(yī)院建筑自身用能特點(diǎn),提出合適的用能評(píng)價(jià)及分析方法。首先,建立了適用于醫(yī)院建筑末端用戶的用能評(píng)價(jià)方法。該方法由綜合用能評(píng)價(jià)、分類分項(xiàng)用能評(píng)價(jià)組成,這兩部分評(píng)價(jià)都是以k-means聚類算法作為基本方法。通過(guò)綜合評(píng)價(jià),不僅可以評(píng)價(jià)用戶總能耗量的高低,還可發(fā)現(xiàn)用戶用能行為的差異,從而為醫(yī)院開(kāi)展用能評(píng)比工作提供依據(jù);通過(guò)分類分項(xiàng)評(píng)價(jià),可發(fā)現(xiàn)用戶的節(jié)能潛力項(xiàng),從而為用戶提供行為節(jié)能方向。其次,提出了醫(yī)院建筑用能分析的具體方法。在用能規(guī)律分析中,總結(jié)了晝夜能耗比、能耗均衡率等8個(gè)用能特性參數(shù),以反映能耗曲線的基本情況。在節(jié)能潛力分析中,提出了兩種計(jì)算方法:一種是與自身歷史情況縱向比較的節(jié)能潛力,另一種是與最優(yōu)用戶橫向比較的節(jié)能潛力。在用能異常情況分析中,提出了兩種分析方法:第一種方法是通過(guò)聚類方法發(fā)現(xiàn)用能異常模式;第二種方法是借助神經(jīng)網(wǎng)絡(luò)對(duì)歷史經(jīng)驗(yàn)進(jìn)行學(xué)習(xí),從而掌握對(duì)異常原因的判斷能力。最后,將所提出的用能評(píng)價(jià)及分析方法,應(yīng)用于某醫(yī)院的實(shí)際案例分析中。對(duì)該醫(yī)院2014年2月的科室能耗進(jìn)行評(píng)價(jià),發(fā)現(xiàn)了綜合用能偏高的23個(gè)科室,并為其提供了具體的行為節(jié)能方向;此外,基于用能評(píng)價(jià)結(jié)果,計(jì)算了各個(gè)科室該月份的節(jié)電潛力。對(duì)醫(yī)院三號(hào)消毒站2013年11月27日的蒸汽用量進(jìn)行能耗規(guī)律分析,并發(fā)現(xiàn)其存在峰谷調(diào)節(jié)潛力。通過(guò)基于聚類的異常分析方法,挖掘出6種典型的用能異常模式;通過(guò)基于神經(jīng)網(wǎng)絡(luò)的異常分析方法,給出了對(duì)用能異常原因的正確判斷結(jié)果。
[Abstract]:In the energy consumption measurement of hospital buildings, a large number of energy consumption data are collected and stored, in which there is often a lot of valuable information. On the premise that the statistics of hospital energy consumption has been completed, starting from the collected original data, through the data mining method, according to the characteristics of hospital building energy consumption, this paper puts forward a suitable energy use evaluation and analysis method. First of all, an energy use evaluation method is established, which is suitable for the end users of hospital buildings. This method is composed of comprehensive energy use evaluation and classification sub-energy evaluation. K-means clustering algorithm is used as the basic method in these two parts of the evaluation. Through comprehensive evaluation, we can not only evaluate the total energy consumption of users, but also find the difference of energy consumption behavior of users, thus providing the basis for the hospital to carry out the evaluation of energy use. Through the classification and sub-evaluation, the energy saving potential items of users can be found, so as to provide users with behavioral energy saving direction. Secondly, the concrete method of energy consumption analysis in hospital building is put forward. In the analysis of energy consumption law, eight energy consumption characteristic parameters, such as diurnal energy consumption ratio and energy consumption balance rate, are summarized to reflect the basic situation of energy consumption curve. In the analysis of energy saving potential, two calculation methods are put forward: one is the energy saving potential compared vertically with the historical situation, the other is the energy saving potential compared with the optimal user horizontally. In the analysis of energy anomaly, two analysis methods are put forward: the first method is to find the abnormal pattern of energy use by clustering method; The second method is to learn the historical experience with the help of neural network, so as to master the ability to judge the abnormal causes. Finally, the proposed energy evaluation and analysis method is applied to the actual case analysis of a hospital. The energy consumption of the department of the hospital in February 2014 was evaluated, and 23 departments with high comprehensive energy consumption were found, and the specific direction of behavior energy saving was provided. In addition, based on the results of the energy use evaluation, the power saving potential of each department in that month was calculated. The energy consumption of steam consumption in No. 3 disinfection station of hospital on November 27, 2013 was analyzed, and it was found that there was peak and valley regulating potential. Through the anomaly analysis method based on clustering, six typical abnormal patterns of energy use are excavated, and the correct judgment results of the causes of abnormal energy use are given by the anomaly analysis method based on neural network.
【學(xué)位授予單位】：哈爾濱工業(yè)大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：TU111.195

【參考文獻(xiàn)】

相關(guān)期刊論文 前1條

1 劉丹丹;陳啟軍;森一之;木田幸夫;;基于數(shù)據(jù)的建筑能耗分析與建模[J];同濟(jì)大學(xué)學(xué)報(bào)(自然科學(xué)版);2010年12期

相關(guān)碩士學(xué)位論文 前2條

1 張奎;山東省醫(yī)院建筑能源審計(jì)與節(jié)能措施研究[D];山東大學(xué);2011年

2 原兆亮;醫(yī)院建筑節(jié)能潛力分析與節(jié)能途徑研究[D];長(zhǎng)安大學(xué);2012年

，

本文編號(hào)：2492223