本文選題：光伏電站 + Elman神經(jīng)網(wǎng)絡(luò)��； 參考：《河北工業(yè)大學(xué)》2015年碩士論文

【摘要】：由于光伏發(fā)電受地理位置、天氣狀況和外界環(huán)境等條件的影響巨大,導(dǎo)致光伏電站的電能輸出具有明顯的間歇性、隨機(jī)性,也就導(dǎo)致了光伏電站內(nèi)部的光伏組件和儲能系統(tǒng)、光伏電站與大電網(wǎng)之間的功率交換過程復(fù)雜化。光伏電站可以從電力公司購電,也可以售電給電力公司,有償?shù)臑殡娋W(wǎng)提供“削峰填谷”和緊急功率支持等服務(wù)。本文以青海某1MW光伏電站為研究對象,對其能量管理的基本理論和管理策略進(jìn)行了探究,其中包括光伏發(fā)電功率短期預(yù)測和實(shí)時能量管理。準(zhǔn)確的預(yù)知光伏組件在未來某段時間內(nèi)的發(fā)電功率,對光伏電站內(nèi)的光伏組件和儲能系統(tǒng)的最優(yōu)配合、經(jīng)濟(jì)調(diào)度、最優(yōu)潮流等具有著深遠(yuǎn)意義。為此,本文基于Elman神經(jīng)網(wǎng)絡(luò)理論研究光伏發(fā)電功率短期預(yù)測模型。首先,研究了氣象因素(如太陽輻照度、溫度等)與光伏發(fā)電功率的相關(guān)性;其次,搭建了基于Elman神經(jīng)網(wǎng)絡(luò)的光伏發(fā)電功率短期預(yù)測模型,確定模型的輸入層、隱含層、輸出層和承接層的神經(jīng)元數(shù)目,并定量評估了不同天氣類型下的預(yù)測模型的預(yù)測精度。與BP神經(jīng)網(wǎng)絡(luò)算法和NSET算法作對比分析研究,驗(yàn)證本文所采用的預(yù)測模型算法比這兩種算法的預(yù)測精度都高。光伏發(fā)電的出力波動劇烈,不宜獨(dú)立向負(fù)荷供電,需要同其它儲能裝置配合使用。此外,光伏電站并網(wǎng)運(yùn)行改變了系統(tǒng)中的潮流分布,所以需要對光伏單元、儲能系統(tǒng)和大電網(wǎng)之間的能量進(jìn)行管理,實(shí)現(xiàn)光伏電站穩(wěn)定并網(wǎng)、高效經(jīng)濟(jì)運(yùn)行。針對光伏單元在并網(wǎng)運(yùn)行中面臨的能量管理問題,本文建立了一種并網(wǎng)光伏電站實(shí)時能量管理模型。首先,從凌晨0:00到24:00劃分為峰、平、谷三個時段;然后隨時跟蹤儲能蓄電池的荷電狀態(tài)SOC,根據(jù)當(dāng)前時刻所處在的不同時段和蓄電池的SOC情況采用不同的能量調(diào)度模型;同時需要考慮儲能蓄電池及其配套裝置成本等;最后通過算例驗(yàn)證了本文所提出的方法不僅可以實(shí)現(xiàn)光伏電站的經(jīng)濟(jì)運(yùn)行,還可以輔助大電網(wǎng)進(jìn)行“削峰填谷”。
[Abstract]:Because photovoltaic power generation is greatly affected by geographical location, weather conditions and external environment, the power output of photovoltaic power station has obvious intermittence and randomness, which leads to the photovoltaic module and energy storage system inside the photovoltaic power station. The process of power exchange between photovoltaic power plants and large power grids is complicated. Photovoltaic power stations can buy electricity from power companies or sell electricity to power companies, providing services such as "peak cutting and valley filling" and emergency power support for the grid. In this paper, a 1MW photovoltaic power plant in Qinghai Province is studied. The basic theory and management strategy of energy management are discussed, including short-term prediction of photovoltaic power generation and real-time energy management. Accurately predicting the generation power of PV module in a certain period of time is of great significance to the optimal coordination, economic dispatch and optimal power flow of PV module and energy storage system in photovoltaic power plant. Therefore, based on Elman neural network theory, this paper studies the short-term prediction model of photovoltaic power generation. Firstly, the correlation between meteorological factors (such as solar irradiance, temperature, etc.) and photovoltaic power generation is studied. Secondly, the short-term prediction model of photovoltaic power generation based on Elman neural network is built to determine the input layer and hidden layer of the model. The number of neurons in the output layer and the receiving layer and the prediction accuracy of the prediction models under different weather types are quantitatively evaluated. Compared with BP neural network algorithm and NSET algorithm, it is verified that the prediction model algorithm used in this paper is more accurate than these two algorithms. The output force of photovoltaic power generation fluctuates sharply, so it is not suitable to supply power to load independently, so it is necessary to cooperate with other energy storage devices. In addition, the grid-connected operation of photovoltaic power station changes the distribution of power flow in the system, so it is necessary to manage the energy between photovoltaic unit, energy storage system and large power grid to realize stable grid connection and efficient and economical operation of photovoltaic power station. Aiming at the problem of energy management in grid-connected operation of photovoltaic unit, a real-time energy management model of grid-connected photovoltaic power station is established in this paper. First of all, from 0:00 to 24:00, it is divided into three periods: peak, level and valley, and then it tracks the state of the storage battery at any time, and adopts different energy scheduling models according to the different periods of time and the SOC of the battery at present. At the same time, it is necessary to consider the cost of energy storage battery and its supporting equipment. Finally, it is verified that the proposed method can not only realize the economic operation of photovoltaic power station, but also assist the large power network to "cut the peak and fill the valley".
【學(xué)位授予單位】：河北工業(yè)大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2015
【分類號】：TM615

【參考文獻(xiàn)】

相關(guān)期刊論文 前1條

1 李光明;劉祖明;何京鴻;趙恒利;張樹明;;基于多元線性回歸模型的并網(wǎng)光伏發(fā)電系統(tǒng)發(fā)電量預(yù)測研究[J];現(xiàn)代電力;2011年02期

，

本文編號：2111181