電力市場環(huán)境下的微電網(wǎng)能量管理系統(tǒng)及風險管理

1、學號：1512302016 姓名：羅任志 專業(yè)：電力系統(tǒng)及其自動化 日期：2016年7月A Microgrid Energy Management System and Risk Management under an Electricity Market Environment電力市場環(huán)境下的微電網(wǎng)能量管理系統(tǒng)及風險管理Jingshuang Shen, Chuanwen Jiang, Yangyang Liu and Xu Wang AbstractThis paper presents a novel energy-management method for a microgrid tha

2、t includes renewable energy, diesel generators, battery storage, and various loads. We assume that the microgrid takes part in a pool market and responds actively to the electricity price to maximize its profit by scheduling its controllable resources. To address various uncertainties, a risk-constr

3、ained scenario-based stochastic programming framework is proposed using the conditional value at risk method. The designed model is solved by two levels of stochastic optimization methods. One level of optimization is to submit optimal hourly bids to the day-ahead market under the forecast data. The

4、 other level of optimization is to determine the optimal scheduling using the scenario-based stochastic data of the uncertain resources. The proposed energy management system is not only beneficial for the microgrid and customers but also applies the microgrid aggregator and virtual power plant (VPP

5、). The results are shown to prove the validity of the proposed framework. 摘要-本文為微電網(wǎng)提出了一種新的能源管理方法，微電網(wǎng)中包括可再生能源，柴油發(fā)電機，電池存儲和各種負載。我們假設，微電網(wǎng)為合伙經(jīng)營市場中的一部分，并通過調(diào)度其可控的資源積極響應的電力價格，以最大限度地提高利潤。為了解決各種不確定性，提出了一種基于場景的風險約束的隨機規(guī)劃的框架，用于條件值的風險評估。設計的模型是由兩個層次的隨機優(yōu)化方法組成。優(yōu)化的一個方面是提交在預測數(shù)據(jù)下日前市場每小時最佳出價。優(yōu)化另一個方面是基于不確定資源的場景的隨機數(shù)據(jù)確定最優(yōu)調(diào)

6、度。本文所提出的能量管理系統(tǒng)不僅有利于微網(wǎng)和客戶也適用于微電網(wǎng)的聚合和虛擬電廠（VPP）。所示結果證明了提出的框架的有效性。 KeywordsControllable load; Smart grid; Energy management; Electricity market; Microgrid; Renewable energy; Risk management; Stochastic optimization關鍵詞：可控負荷；智能電網(wǎng)；能源管理；電力市場；微電網(wǎng)；可再生能源；風險管理；隨機優(yōu)化I. INTRODUCTIONI. 簡介In recent years, the microg

7、rid has been growing dramatically due to its potential benefits to provide reliable, secure, efficient, environmentally friendly, and sustainable electricity from renewable energy sources12. A microgrid consists of distributed energy sources, such as micro turbines, wind turbines, fuel cells and pho

8、tovoltaic system (PVs), storage devices and a group of radial load feeders3. In the grid-connected mode, a microgrid is connected to the grid through a point of common coupling in a low-voltage distribution network. With the development of smart grid technologies, more and more controllable resource

9、s in the microgrid can exchange information with the higher-level power system45. Hence a microgrid can respond actively to the electricity price to maximize its profit by scheduling its controllable resources. In 6, a price-incentive model was utilized to generate a management strategy to coordinat

10、e the charging of electric vehicles (EVs) and battery swap stations (BSSs) to minimize the total cost of the EVs and maximize the profit from the BSS in grid-connected mode.In 7, the author designed an objective to determine the optimal hourly bids that the microgrid aggregator should submit to the

11、day-ahead market to maximize its profit. In 8, two market bidding techniques, single bidding and discriminatory bidding, were proposed for the microgrid to participate in the bidding process. Much work has been carried out on bidding and auction theory in the competitive electricity market. However,

12、 the energy management and optimal operation for the microgrid under the electricity market environment face challenges.近年來，微電網(wǎng)由于其潛在的好處有了極大的發(fā)展，其能提供可靠，安全，高效，環(huán)保，可再生能源和可持續(xù)的電力能源 1 2 。一個微電網(wǎng)的分布式能源，如微型燃氣輪機、風力發(fā)電機、燃料電池、光伏發(fā)電系統(tǒng)（PVS），存儲設備和一組徑向載荷饋線 3 。在并網(wǎng)模式下，一個微電網(wǎng)通過在低壓配電網(wǎng)中的公共耦合點連接到電網(wǎng)。隨著智能電網(wǎng)技術的發(fā)展， 微電網(wǎng)中越來越多的可控能源可以

13、與高電壓的電力系統(tǒng)進行信息交換 4 5 。因此，一個微電網(wǎng)可以積極響應其電力價格并通過可控資源的調(diào)度，以最大限度地提高其利潤。在 6 ，利用價格激勵模型來產(chǎn)生一個管理策略來協(xié)調(diào)充電的電動車（EV）和電池交換站（BSSS），以降低電動汽車的總成本和最大限度提高來自在并網(wǎng)中的電池交換站的利潤。在圖 7 中，作者設計了一個服從日前市場利潤最大化的每小時最佳出價的目標函數(shù)。在 8 中，為微電網(wǎng)的參與競價提出了兩市場招投標技術、單投標和有偏見投標技術。競爭的電力市場中的招標和拍賣已經(jīng)完成了許多工作。但是，在電力市場環(huán)境下微電網(wǎng)的能源管理和優(yōu)化運行面臨挑戰(zhàn)。As it is an important re

14、search field of smart power, several approaches have been reported in the literature in relation to microgrid smart energy management applicable within the smart grid system. In 9, the authors proposed multi-objective intelligent energy management to minimize the operation cost and the environmental

15、 impact of a microgrid. In 10, a novel double-layer coordinated control approach for microgrid energy management was proposed, including a schedule layer obtaining an economic operation scheme based on forecasting data and a dispatch layer providing power to controllable units based on real-time dat

16、a. In 11, three-level hierarchical energy management strategies were presented for multi-microgrids. Demand response and demand side management have also been considered in the microgrid energy management system 12 13. Overall, most existing works have not roundly considered the uncertainties of ele

17、ments in the microgrid system. Renewable energy, such as wind and photovoltaic generation, customer loads and market electricity prices are uncertain and random in real time. Although some works considered the uncertainties of renewable energy 1011, the uncertainties of market electricity prices hav

18、e seldom been considered.在許多與適用于智能電網(wǎng)系統(tǒng)的微電網(wǎng)智能能源管理相關的文獻中已經(jīng)提出，這是智能能源中一個重要的研究領域。在 9 ，作者提出了多目標智能能源管理，以盡量減少微電網(wǎng)的運行成本和對環(huán)境的影響。在 10 ，提出了使用一種新型的雙層協(xié)調(diào)控制的方法以達到微電網(wǎng)能量管理，其中包括一個基于預測數(shù)據(jù)獲得經(jīng)濟運行的計劃層和一個基于實時數(shù)據(jù)對單元進行功率可控的調(diào)度層。在 11 ，為多微網(wǎng)提出了三層次的能量管理策略。需求響應和需求側(cè)管理也被考慮在微電網(wǎng)能量管理系統(tǒng)中 12 13 ?？傮w而言，大多數(shù)現(xiàn)有的工作沒有全面考慮在微電網(wǎng)系統(tǒng)中的不確定元素?？稍偕茉?，如風和 光伏

19、發(fā)電，客戶負載和市場電價是不確定的，實時為隨機的。雖然有些工作可以視為考慮了可再生能源的 10 11 的不確定性，市場電價的不確定性很少被考慮。In this paper, we present a microgrid energy management system that considers the uncertainties of renewable resources, customer loads and electricity prices. To address various uncertainties, we propose a double-layer scenario-

20、based stochastic optimization approach. The first layer obtains an economic operation scheme based on forecasting data, while the second layer provides the power to controllable units based on real-time data. The microgrid schedules the controllable resources to maximize its profit. However, the pro

21、fit may be at risk due to the uncertain resources in scenario-based stochastic programs. To constrain the risk, risk management is also proposed in the objective function using the conditional value at risk method.在本文中，我們提出了一種考慮了可再生資源，客戶負載和電力價格中的各種不確定性的微電網(wǎng)能源管理系統(tǒng)。為了解決各種不確定性，本文提出了一種基于場景隨機的雙層優(yōu)化方法。第一層獲得

22、基于預測數(shù)據(jù)的經(jīng)濟運行方案，而第二層基于實時數(shù)據(jù)對功率可控單元進行控制。微電網(wǎng)調(diào)度可控的資源以最大限度地提高其利潤。然而，由于基于隨機場景程序中的不確定資源的存在，利潤是有風險的。為了約束風險，在使用了條件值風險法的目標函數(shù)中也提出了風險管理的方法。The remainder of this paper is organized as follows. Section II describes the system model. Section III describes the solution approach. A detailed problem formulation is pres

23、ented in Section IV. Several case studies and numerical results are provided in Section V. Finally, Section VI states the concluding remarks and discusses some directions for future works.本文的其余部分內(nèi)容如下。第二節(jié)介紹了系統(tǒng)的模型。第三節(jié)介紹了解決的方案。第四節(jié)描述了一個詳細問題的。第五節(jié)闡述了幾個案例的研究和數(shù)值結果。最后，第六節(jié)是本文的結語和討論今后的工作方向。II. MICROGRID COMPON

24、ENT MODELING II.微電網(wǎng)構件建模A.Price ModelingA.價格模型We assume that a microgrid is connected to the main grid and the grid supplies power to the microgrid to balance the microgrid demand. A two-way communication network is available for a microgrid management center to control the controllable units. We ass

25、ume that the microgrid possesses a few diesel generators, storage batteries, wind turbines, PV panels, and controllable loads. The microgrid can procure energy from the wholesale electricity market and can also sell energy back to the market when the local generation is surplus. Under the electric p

26、ower pool mode, the microgrid is a price-taker. It submits the hourly power quantities that it commits to buy/sell in the day-ahead (DA) energy market to the market operator before the operating day. During the operating day, the microgrid participates in the real-time energy market to compensate fo

27、r the deviation from the day-ahead schedule. The market electricity price and quantity are formulated as 我們假設有一個微電網(wǎng)連接到主電網(wǎng)，主電網(wǎng)向微電網(wǎng)提供電能以平衡微電網(wǎng)的需求。一個雙向通信網(wǎng)絡應用于微電網(wǎng)的控制管理中心進行控制單元。我們假設這個微電網(wǎng)里包含了一些柴油發(fā)電機，存儲電池，風力渦輪機，光伏電池板，和可控負載。微電網(wǎng)可以從批發(fā)電力市場購買電能，而產(chǎn)能過剩時也可以將電能賣回市場。在電力聯(lián)營模式下，微電網(wǎng)是價格接受方。其每小時提供的電量為在工作日之前承諾購買或出售給能源市場的數(shù)量。

28、在工作期間，微電網(wǎng)參與到實時能源市場中以彌補其錯誤前一天的進度偏差。市場電價和數(shù)量的公式如下：whererepresents the market electricity price, and represent the forecast electricity price and forecast error, respectively, represents the electricity quantity that the microgrid buys from the electricity market, and and represent the planning purchase

29、 quantity and real-time variation, respectively. The forecast error includes the active variation and passive variation. The active variation represents the variation that the microgrid dispatches actively to maximize its profit when the supply is abundant within the microgrid. The passive variation

30、 represents the variation by which the microgrid must purchase energy from the main grid when the supply is not enough due to forecast error. The mathematical formula is 代表市場電價，和分別代表預測電價和預測誤差，代表從微電網(wǎng)購買電力市場的電量，和分別表了計劃采購量和實時變化量。預測誤差包括主動變化 和被動變化。主動變化指的是，當微電網(wǎng)調(diào)度積極追求利潤最大化時對微電網(wǎng)調(diào)度的變化。被動變化指的是，微電網(wǎng)由于預測誤差而供應不夠時必

31、須從主電網(wǎng)中購買電量的變化。數(shù)學公式是： where and represent the active variation and passive variation, respectively.和分別代表了主動變化和被動變化。B.Demand Response需求響應We assume that the controllable loads are effective controllable units that respond actively to the electricity price. There are two types of controllable loads with

32、in the microgrid. One type of is passive such as refrigerators, freezers, air conditioners, water heaters and heat pumps, which can be controlled by direct load control (DLC) and interruptible load management (ILM).The other type is active, such as vehicle-to-grid(V2G) and heat storage, which not on

33、ly can be controlled like the first type but can also supply energy to the main grid. This type can more effectively take part in load management programs. In the electricity market, the controllable loads respond actively to the electricity price as我們假設，可控負載是有效的可控單元并能積極響應電力價格。微電網(wǎng)中有兩種類型的可控負載。一種是被動類型

34、例如冰箱、冰柜、空調(diào)、熱水器和熱泵，它們可以通過直接負荷控制（DLC）和可中斷負荷管理（ILM）進行控制。另一種是主動類型，如車輛接入電網(wǎng)（V2G）和儲熱的，它們不僅可以像第一類一樣進行控制而且還可以向電網(wǎng)提供能量。這種類型可以更有效地參與負載管理程序中。在電力市場中，可控負載可以積極響應電價：Where and represent the electricity price in real time and the reference price, respectively, and represent controllable loads under and respectively, a

35、nd k represents the controllable loads are also stochastic.和分別代表了實時電價和參考電價，和分別代表在和下的可控負荷，K 表示可控負載的隨機系數(shù)。C.Renewable EnergyC.可再生能源Renewable resource generation such as wind turbines and PV panels is uncertain. For example, the output of wind turbines depends on the wind speed, and the output of PV pan

36、els depends on the irradiance and temperature.可再生資源發(fā)電，如風力渦輪機和光伏板是不確定的。例如，風力渦輪機的輸出取決于風速和光伏電池板的輸出取決于輻射和溫度。1)Wind Turbines1)風力渦輪機The output power of the wind turbines is described by :風力發(fā)電機的輸出功率描述為：where and represent the output power and rated output power of the wind turbine, respectively; , , and re

37、present the wind speed, cut-in speed, rated speed and cut-off speed of the wind turbine, respectively; and a and b are fitting parameters of the wind turbine power curve.和分別代表了風力發(fā)電機的輸出功率和額定輸出功率，、分別代表風力發(fā)電機的風速、轉(zhuǎn)速、額定轉(zhuǎn)速和截止轉(zhuǎn)速，a和b是風力發(fā)電機功率曲線的擬合參數(shù)。As shown in Eq. (5), the output of the wind turbines is depe

38、ndent on the wind speed, which is obtained by the forecast in the energy-management system, but is unpredictable.如式（5）所示，風力發(fā)電機的輸出依賴于風的速度，這是由能源管理系統(tǒng)預測所得到的，但不是可預測的。2)PV panels2)光伏電池板The output of a PV generator is a function of the irradiance and temperature, which is provided by a confirmed formula,光伏

39、發(fā)電機的輸出是輻射和溫度的函數(shù)，它由一個確定的公式構成，where represents the maximum output under standard test conditions; and represent the current irradiance and standard irradiance, respectively; and represent the current temperature and standard temperature, respectively; and k is a temperature coefficient.表示標準測試條件下的最大輸出；

40、和分別代表當前輻射和標準輻射，和 分別代表了當前溫度和標準溫度，K為一個溫度系數(shù)。III. MODELING APPROACH 建模的方法A.Stochastic Optimization ApproachA.隨機優(yōu)化方法In this paper, we propose a two-stage scenario-based stochastic programming approach to address the uncertainties in the microgrid. In the first stage, the forecast data of the uncertaintie

41、s such as the wind speed, PV power, loads and electricity prices can be obtained by traditional forecasting techniques. Then, a Monte Carlo simulation with the Latin hypercube sampling technique is implemented to generate a large number of scenarios representing values of the uncertain parameters. F

42、orecasting errors are always present. For simplicity, the forecasting errors of the wind speed, PV power, loads and electricity prices are assumed to follow normal distributions in this paper.在本文中，我們提出了一個兩階段的基于場景的隨機規(guī)劃方法，以解決在微電網(wǎng)中存在的不確定性。在第一階段中，不確定性數(shù)據(jù)的預測，如風速、光伏發(fā)電、負荷和電價可以由傳統(tǒng)的預測方法獲得。然后，用帶有Latin hypercub

43、e采樣技術的Monte Carlo模型實現(xiàn)生成大量場景的不確定參數(shù)值。預測錯誤總是存在的。為簡單起見，在本文中風速，光伏發(fā)電，負載的預測誤差和電力價格被假定為遵循正態(tài)分布的 。It is desirable to generate a large number of scenarios to increase the accuracy of the results. However, the number of generated scenarios directly impacts the computational complexity. To address this trade off

44、, 2000 scenarios are generated using a Monte Carlo simulation with the Latin hypercube sampling(LHS) technique in this paper. A fast-forward reduction method such as The general algebraic modeling system (GAMS)/ scenario reduction (SCENRED) is implemented to reduce the computation time from 2000 sce

45、narios to 200 scenarios without affecting the accuracy of the optimization results.它通過產(chǎn)生大量的情況以提高結果的準確性，結果是理想的。然而，所產(chǎn)生的情況直接影響了計算的復雜度。為了達到權衡，本文使用了帶有Latin hypercube 采樣技術的Monte Carlo模擬（LHS）生成技術來產(chǎn)生2000種情況。一種快速的前向還原方法，如在不影響優(yōu)化結果的準確性下，用一般代數(shù)建模系統(tǒng)（GAMS）/場景還原（SCENRED）來減少計算從2000到200種情況下的時間。B.Risk ManagementB.風險管理

46、As data such as the wind speed, PV power, loads and electricity prices are produced randomly in the scenario-based stochastic optimization programming, the profit of the microgrid in the proposed model is indeed uncertain. The optimal expected profit under some scenarios may be very low or even nega

47、tive. As a result, the expected profit may be variable and faces a high level of risk. In this paper, we propose a risk management scheme, namely, conditional value at risk (CVaR), to control the trade-off between the expected profit and its variability. 由于如風速，光伏發(fā)電，負載和電力價格這些數(shù)據(jù)在基于場景的隨機優(yōu)化規(guī)劃中是隨機產(chǎn)生的，所提出

48、的微電網(wǎng)模型的利潤的確是不確定的。在某些情況下，最佳的預期利潤可能是非常低的，甚至是負的。因此，預期的利潤可能是可變的，并面臨著高水平的風險。在本文中， 我們提出了一個風險管理方案，即條件風險價值（CVaR），用以控制預期利潤與其變化之間的權衡。Several risk measures have been introduced to quantify risk 14, and one of the most popular is Value-at-Risk (VaR). However, it has undesirable mathematical characteristics such

49、 as a lack of subadditivity and convexity. VaR is also difficult to optimize when it is calculated from scenarios. As an alternative measure of risk, CVaR is known to have advantages over VaR in that it is transition-equivariant, positively homogeneous, convex, and has a stochastic dominance of orde

50、r 1. In the scenario-based stochastic optimization method, the conditional value at risk at the confidence level (-CVaR) can be defined as the expected profit in the (1-)100% worst scenarios,which is expressed as 14幾個風險措施已被引入到量化風險中 14 ，其中最受歡迎的是風險價值（風險值）。然而，它有不良的數(shù)學特征，例如缺乏可加性和凸性。當從多個場景來計算是是風險值也是難以優(yōu)化的。

51、作為風險的替代計量，CVaR對于VAR具有過渡等變，正齊次，凹凸有致，1階隨機主導地位的優(yōu)點。在基于場景的隨機優(yōu)化方法中，風險條件值的置信水平（CVaR）可以被定義為在在（1 -）100%最壞的情況的預期利潤，可以表示為式 14 Where re presents a random variable and represents a confidence level.re表示一個隨機變量，表示置信水平。In 7, the author supplies a CVaR solving method, which is represented as ：在 7 ，作者提出了CVaR的求解方法，可以表

52、示為：where NS represents the number of Monte Carlo scenarios; s represents the probability of scenario s, represents the VaR, and represents an auxiliary nonnegative variable equal to the difference between the VaR and profits is smaller than VaR and equal to zero otherwise.其中n代表Monte Carlo情況的數(shù)量；代表某種情

53、況的概率，代表無功，和表示一個輔助的非負變量等于VaR和利潤之間的差大于VaR的體積更小，等于否則為0。IV. P ROBLEM F ORMULATIONIV.問題描述In this paper, we propose a double-level stochastic optimization method to maximize the profit of the microgrid. We assume that the microgrid always operates in grid-connected mode. The objective of the model is to m

54、aximize the profit of the microgrid over a given time period together with achieving risk management. Normally, the operation costs of renewable energy and energy storage such as batteries are minimal. Therefore, renewable energy and battery operation costs are not considered in this paper. The obje

55、ctive function is given as在本文中，我們提出了一個雙層隨機優(yōu)化方法，以最大限度地提高微電網(wǎng)的利潤。我們假設，微電網(wǎng)總是運行在連接電網(wǎng)的模式。該模型的目的是給定的時間段實現(xiàn)微電網(wǎng)的風險管理使利潤最大化。通常情況下，可再生能源和能源存儲運行成本的是最小的，例如電池。因此，本文不考慮可再生能源和電池的運行成本。目標函數(shù)可以表示為：where N S represents the number of Monte Carlo scenarios; s represents the probability of scenario s; profits represents the

56、 profit of the microgrid in scenario s; and represents the risk aversion parameter. When is equal to zero, the microgrid is a risk-neutral decision maker. With increasing, the microgrid becomes more risk-averse其中n代表Monte Carlo情況的數(shù)量；代表情況的概率； s代表微電網(wǎng)在某種情況下的利潤；代表風險規(guī)避參數(shù).當?shù)扔诹?，微網(wǎng)是一個風險中性的決定者。隨著增加，微網(wǎng)變得越來越規(guī)避風

57、險。The profit of the microgrid in scenarios is在某種情況下微電網(wǎng)的利潤是：where represents the profit of the microgrid obtained from the main grid; represents the cost of the distributed generation (DG) in the microgrid; and represent the electrical energy sold and bought from the grid tie line in scenario s, resp

58、ectively; and represent the sell and buy electricity prices in scenario s, respectively; and represent the penalty amount and penalty price when the actual power is not equal to the plan exchange power in the tie line and scenario s;represents the i th generator cost of fuel; represents the i th generator maintenance cost; represents the total number of diesel generators; represents the i generator power in period t and scenario s; represents the i generator start-up cost; and represents the i generator status in perio