2015 IEEE PES Innovative Smart Grid Technologies Latin America (ISGT LATAM)

A Proposal for Microgrids Control Architecture as

Aggregator
Alonso de Jesús Chica Leal César Trujillo Rodrı́guez Francisco Santamarı́a Piedrahita
Universidad Central Universidad Distrital FJDC Universidad Distrital FJDC
Universidad Distrital FJDC Bogotá, Colombia Bogotá, Colombia
Bogotá, Colombia Email: [email protected] Email: [email protected]
Email: [email protected]

Abstract—This paper presents a review of microgrid control gator in demand response. The possibility of aggregated MGs
architectures, and it proposes a decentralized control architecture having a contract with the distribution grid operator to supply
for multiple microgrids operating as a power aggregator in energy for demand response is analyzed.
demand response. This architecture consists of four control levels.
The interrelationships of the hierarchical levels in the proposed In Section II, a review of microgrid control architectures
architecture are described, and the operation models of control is presented. In Section III, the concept of power aggregator
architectures for each microgrid and for multiple microgrids using MGs is explained. Proposed MGs control architecture
operating as aggregator are presented. Finally, some comments including a power aggregator level is presented in Section IV.
about the proposal are presented. Finally, in Section V some comments about the proposal are
presented.
I. I NTRODUCTION
Distributed Energy Resources (DER) for Demand Response II. R EVIEW OF MICROGRID CONTROL ARCHITECTURES
(DR) are essential components in the future of smart grids Depending on the control architecture, it is necessary to have
[1], seeking to integrate Renewable Energy Resources (RER) the set points available to local controllers of power generators,
to keep the balance between energy demand and supply. storage devices and smart loads, as well as the island and grid
Microgrid (MG) is a subsystem of the distribution grid, which operation modes [12]. There are mainly three MG control
comprises generation capacities, storage devices, and control- architectures: centralized, decentralized and multilayer. The
lable loads, operating as a single controllable system either last architecture is also called hierarchical control levels [13]-
connected or isolated from the utility grid [2]–[4]. Further- [16].
more, as a group of DER, MGs can provide flexibility to the In centralized control, architecture sources are controlled
distribution system to get a faster response to energy demand, by a central control system, located remotely, in a Mater-
and improve the capacity to reduce distribution network losses Slave configuration as depicted in figure 1. In this architecture,
[3],[5]. the Microgrid Central Controller (MCC) optimizes power
Microgrids can be AC, DC, AC/DC, or high frequency [3]– interchanges between MG and the main grid, which maximizes
[5] and can operate in both modes, island mode and grid- the local generation, depending on security and market prices
connected mode [6]–[8]. The requirements for each operation restrictions. This is possible with control points to DER and
mode are different as well as the specifications for stability controllable loads into MGs.
and control [4]. MGs are controlled systems that can operate In this architecture the use of bidirectional communication
as a load or an aggregated generator, and they can act as power networks between MCC and each Local Controller (LC) is
sources incorporated into the network or as a means to provide necessary. This communication can be done by using telephone
ancillary services [9], [10]. lines, power line communication, serial lines, or wireless
In a distribution system, hundreds of DER units can be communication [17], [18], being Serial and TCP Modbus and
available to propose an aggregated load to the main grid. This Control Area Network (CAN) the most used [19].
leads to a smart distribution system that requires real-time In decentralized control, the individual sources share the
information about each DER unit and load [11]. On the other load according to their characteristics and capacities of individ-
hand, the complexity problem of control is reduced to control ual power control without any communication between them.
all DER units for MGs dispatch. In this sense, MGs are smart In this concept, most drivers follow a droop control scheme,
systems that can respond to energy demand by increasing their either for operation in island mode or grid connected mode
local generation or by turning off non-critical loads or shifting [20]-[22]. Figure 2 depicts the independence of the drivers
their use in time. and the absence of a central controller. Real–time economic
This paper presents a proposal for a decentralized control dispatch optimization and the use of sources under several
architecture for multiple microgrids, acting like a power aggre- operation conditions are not possible using this approach.

978-1-4673-6605-2/15/$31.00 ©2015 IEEE 509

 2015 IEEE PES Innovative Smart Grid Technologies Latin America (ISGT LATAM)

Fig. 1: Centralized control architecture, modified from [15]

Fig. 3: Supervisory control

III. M ICROGRIDS AS POWER AGGREGATORS

MGs in grid-connected operating mode can work as De-
mand Response Aggregators (DRA), and multiple grouped
MGs can take part as client of Demand Side Management
Fig. 2: Decentralized control architecture, modified from [14] (DSM) [25]. However, DER in MGs are small loads with low
capacity, and in most cases, they are not qualified for DSM.
Therefore, it is necessary to check their role as DRA, analyzed
The multilayer control or hierarchical control, combines the as a Virtual Power Plant (VPP), using a business model for
best characteristics of the centralized and decentralized control DRA, allowing them to participate in the wholesale energy
architectures [14]. This concept is used in MGs from experi- market as flexible entities of DR. Multiple grouped MGs can
ence of hierarchical control of dispatch for large-scale power operate as a great elastic load participating in DSM.
systems, where voltage amplitude and frequency deviations are When MGs have non-dispatchable generators [26] such as
limited, and therefore system quality, stability and reliability photovoltaic and wind generators, or others using intermittent
are improved [19]. In [22], a level of control is included to resources [30], [31], it is necessary to adapt the distribution
correct these deviations. Moreover, a control strategy without system to provide flexible and effective response to MGs oper-
communication networks is proposed, which could be appro- ation, ensuring power system reliability under the penetration
priated for high-performance paralleled inverters [23]. of renewable energies and the growing of the temporal loads,
Supervisory control in MGs is proposed in [15], by means of as electric vehicles [32].
three hierarchical control levels. In the top level or grid level, The current and future problem is associated to the orga-
there are both, a controller of Distribution Network Operator nization of the dynamics of the energy markets, based on the
(DNO) and a Market Operator (MO). At the management level, next MGs in the rural region, cities, or near to cities. The main
there is a MCC for each MG and, at the field level or bottom competitive advantage of the multiple grouped MGs is that it
level, LC of DER and the controlled loads are working for can quickly provide electric power to the main grid at a lower
each MCC (See figure 3). price than the the main grid spot market price [30]. Therefore,
Depending on MG control architecture, LC has a different in the future, the organization of the energy market strategies
intelligence level. In centralized operation, each LC receives must be based on MGs.
the set points of the corresponding MCC. In a decentralized According to [31], an aggregator is an entity that collects
operation, each LC makes a decision locally. generated energy for by one or several Distributed Generators
The discussion on problems of microgrid controls include (DG) and sells it to the utility [34]. Additionally, the DRA
multiples generators and energy storage, compensation of can get benefit by participating in pricing between demand
power quality problems and the inclusion of AC, DC, AC/DC and supply.
MGs, among others. Advanced control strategies, such as If a system operator requires a certain amount of power from
distributed and cooperative strategies have been considered. a single or several DRAs at a minimum purchase price, a two
Several proposals involve multi- agents systems with local stages communication from DRA begins. The first stage is to
agents for local tasks and leading agents to coordinate local set price between the system operator and the DRA, and the
agents [24], [25]. other stage is to set the price between the DRA and the MGs

510
 2015 IEEE PES Innovative Smart Grid Technologies Latin America (ISGT LATAM)

[31]. Figure 4 shows a conceptual architecture of several MGs of n Distributed Generators (DG) Gnxy , n energy storage
and a single DRA, where there are n MGs, each one with DER systems ESSxy , and n controllable loads Lnxy with a specific
units and loads operating locally. In addition, communications geographic location xy are considered in the analysis.
between the n MGs and the DRA node are performed, in order DG can be wind, PV or diesel generators, the former ones
to carry out transactions with the wholesale and the retail are non-dispatchable generators and depend of the primary
market, participating in price fixing of aggregated power by energy source (wind and solar radiation), while the last one is
MGs. dispatchable, but with variable output power (1).

Gnxy = {Gnx1 y1 , Gnx2 y2 , . . . , Gnxn yn } (1)

For each Gnxn yn 2 Gnxy , the power complex is:
Sg (t) = Pg (t) + jQg (t) (2)
The set of distributed ESS for any MG is shown in (3)
SAnxy = {SAnx1 y1 , SAnx2 y2 , . . . , SAnxn yn } (3)
For each ESSnxn yn 2 ESSnxy , the State-of-Charge (SOC)
is estimated according to [39]:

SOC(t + k) = SOC(k) ↵(pg (k)) (4)

Fig. 4: Conceptual proposal of MGs as power aggregator Where SOC(k) is the actual ESSnxy , state of charge, Pg (k)
is the power required from DERs in each MG to charge the
Smart transactions between the new load aggregator and ESS, and is the efficiency of charge/discharge of the ESS.
the wholesale and complementary markets haves been already The set of distributed controllable loads in a MG is shown
studied with multi-agent systems [32], [33]. This paper pro- in (5).
poses a solution from the Model Predictive Control (MPC)
Lnxy = {Lnx1 y1 , Lnx2 y2 , . . . , Lnxn yn } (5)
approach.
B. Local control level (LCL)
IV. P ROPOSED CONTROL ARCHITECTURE
At local control level, each DER and controllable loads have
Proposed control architecture is shown in figure 5. In this
a local controller (LC), and there is two-way communication
architecture, DRA is responsible for initiating all the local
with a Central Controller Master (CCM) of each MG, which
markets and making energy auction between the traders of
coordinates the operation of DERs and controllable loads.
different MGs and the main grid.
C. Supervisory Control Level (SCL)
DRA interprets the options of demand response proposed by
load agents, who are monitoring the whole aggregate power
of DER units and the power cost at Power Aggregator Level
(PAL). In this level, PAL receives aggregate cost information
for each MG, estimates the power for each MG aggregation
and sends reference signals to the MGs controllers. Based on
demand predictions and power generation costs, the Model
Based Predictive Control Systems (MBPCS), evaluate MG
participation in the sale of energy [35]. The MBPCS operates
in a decentralized and distributed architecture and it is located
at the Supervisory Control Level (SCL). At this level, control
tasks are performed using the sliding horizon concept to de-
Fig. 5: Proposed architecture for MGs as power aggregator velop predictions in a period determined by MG requirements.
At the supervisory control level, CCM uses a MBPCS
to execute an optimization algorithm to minimize errors of
A. Electrical Power System the SOC estimation of each ESS. It assigns set-points of
In general, low-voltage distribution networks are radial active and reactive power and sends signals for connection
topologies. However, to allow MGs participate as power and disconnection of DERs. The optimization of energy flow
aggregators, more flexible topologies are required, but those between various resources in the MG is developed by the
are not studied in this paper. Therefore, a radial low-voltage MBPCS each hour during the day, using the concept of sliding
distribution grid and n radial MGs, each one with a set horizon.

511
 2015 IEEE PES Innovative Smart Grid Technologies Latin America (ISGT LATAM)

MBPCS has a predictive controller based on model predic- associated with energy production and startup and shutdown
tive control with the following function: decisions, profits and penalties is proposed.
Photovoltaic and wind generators are taken into account in
N Nu the predictive control in each MG, from a forecasting model
X X
J= [ i (SOCi (t + k)) SOCiref ] + [↵i Pi2 (t + k) of energy resources and demand in the MG, and DERs power
k=1 k=1 estimation. In [39] a comparison between load forecasting
+ i Pi2 (t + k)] models using neural networks (NN) and fuzzy inference sys-
(6) tems (FIS) is presented. Short time forecasting models for solar
radiation (from few seconds to several minutes), two states
where N is the prediction horizon and N u is the control model and the ARIMA model, are presented in [40], and other
horizon [40]. The tracking error minimization between the models are available in [41], [42] and [43].
value of SOCi (t + k) and the State-of-Charge of reference In [44] a forecasting model using Multi-Layer Perceptrons
SOCir ef is presented in the horizon N , and error weights (MLPs) and the radial basis function (RBF) NN for wind
i penalize tracking error. On the horizon N u, both the use prediction are described. In [45] relevance vector machines are
of DERs and their power variations are minimized, and those used for wind speed prediction, while reactive power dispatch
are penalized by error weights ↵i and i , respectively. Values is calculated by means of a genetic algorithm. For interested
of N , N u and ↵i , i and i are adjusted according to MG readers, in [46] a complete review of forecasting models for
conditions [37]. solar and wind power is presented.
At each time step, an optimal control problem in open loop
based on measurement and prediction of the input and output E. Power Aggregator Level (PAL)
variables proposed in the objective function is formulated. In DRA interprets the options of demand response proposed by
the optimal solution, just the control action for the current load agents, who are monitoring the whole aggregate power
period is sent to local controllers. This process is repeated at of DER units and the power cost at Power Aggregator Level
each time step, updating both the load-forecasting model and (PAL). In this level, PAL receives aggregate cost information
the short-time forecasting model of the power sources and their for each MG, estimates the power for each MG aggregation
estimated costs. This type of controllers incorporates predictive and sends reference signals to the MGs controllers.
models and restrictions [38], such as a power balance between In this level, the power in each MG is estimated, and
generation and demand, inequality restrictions in the minimum based on that an aggregated power cost optimization process
and maximum power capacity, and the restriction of SOC limit is developed. The total aggregated power and its cost per hour
of storage systems. The optimization problem can be solved are also calculated. This information is sent to the DRA. The
by several methods, where quadratic programming and mixed DRA must know directly the wholesale price of energy to
integer programming are the most widely used. participate in the DR programs, and simultaneously transfer
price signals to end-user, with different risk price protection
D. Energy Management Level (EML)
programs.
In the Energy Management Level (EML) of each MG, DER Power aggregator and DR aggregator take into account the
power and availability are measured, forecasting and prediction MGs participation in the market, for example, following the
models are constructed, and a MG demand prediction is Herfindahl index (H) calculated according to (7), where Si is
developed. the market share of the i–th MG.
At the level of energy management level, the MBPCS in
n
X
each MG defines and solves an optimization problem for
each MG. Aat each time step of time. In this sense, the H= Si2 = S12 + S22 + . . . , Sn2 (7)
i=1
controller must have take high-level decisions about: i) when
to sStarting and stop of each generation unit; ii) how How When H approaches zero, market is more competitive, and
much powerenergy each unit must produce each unit to meet as it approaches one market is almost a monopoly. The use of
cover the load demand at a minimal cost; iii) theThe amount aggregators increases competition and moves the Herfindahl
of energy that has to be stored; iv) when to perform the The index closer to zero. Thus, the active demand response, self-
criteria for charging and discharging ESS cycles of energy sufficiency and selling surplus is encouraged, and there are
storage systemESS; v) If When the MG is connected to the new opportunities with controlled interruptible loads.
network, the quantity when and how much of energy that After performing market transactions with DRA, PAL com-
must be negotiatedbe (bought or sold) from to with the system municates with EML to define the lower reference values,
operator and (when the MG is connected to the network); vi) maintaining a loop between control architecture levels. The
load shedding reduction schedule. MBPCS for PAL and MBPCS for SCL, have transition times
To minimize the functional cost, which represents operat- and different prediction and control horizons, but they are
ing costs, an economic optimization must be performed in synchronized every half hour. The control signals between the
each MG. Therefore, a cost function that includes the costs SCL and the LCL are reported in seconds

512
 2015 IEEE PES Innovative Smart Grid Technologies Latin America (ISGT LATAM)

V. F INAL COMMENTS [16] A.G. Tsikalakis and N.D. Hatziargyriou, Centralized control for opti-
mizing microgrids operation, IEEE Trans. Energy Convers., vol. 23, pp.
Each MG can be dispatched whether locally or can be 241248, 2008.
grouped with other MGs, to participate in DR market using [17] B.A. Vaccaro, M. Popov, D. Villacci, and V. Terzija, An Integrated
Framework for Smart Microgrids Modeling, Communication, and Ver-
power aggregators. Transactions between MGs, DRA and grid ification, Proc. IEEE, vol. 99, no. 1, pp. 119132, 2011.
operator, depend on the costs and demand predictions. [18] R.H. Lasseter, MicroGrids, Power Engineering Society Winter Meeting,
MBPCS allows planning demand due to price variations or 2002. IEEE, vol.1, pp.305,308, 2002.
[19] C.K. Sao and P.W. Lehn, Control and power management of converter
local generation. The proposal shows not only a way of starting fed microgrids, IEEE Trans. Power Syst., vol. 23, no. 3, pp. 10881098,
a new market for existing MGs, but also enables the design of 2008.
new MGs with existing or new DERs, as a form of business [20] J. Rocabert, A. Luna, F. Blaabjerg, and P. Rodrguez, Control of Power
Converters in AC Microgrids, Power Electronics, IEEE Transactions on,
in energy markets for isolated areas. vol. 27, no. 11, pp. 47344749, 2012.
Developing and consolidating a MGs market is not easy [21] M. Hua, H. Hu, Y. Xing, and J.M. Guerrero, Multilayer control for
in the current regulatory, technical and financial framework inverters in parallel operation without intercommunications, IEEE Trans.
Power Electron., vol. 27, no. 8, pp. 36513663, 2012.
[51] in countries like Colombia, where most attempts have [22] J.M. Guerrero, J.C. Vasquez, J. Matas, L.G. De Vicua, and M. Castilla,
been academic experiments. Therefore, further work on this Hierarchical control of droop-controlled AC and DC microgrids - A
approach is necessary. general approach toward standardization, IEEE Trans. Ind. Electron.,
vol. 58, no. 1, pp. 158172, 2011.
[23] J.M. Guerrero and J. Miret, Inverters in Microgrids Using Resistive
R EFERENCES Output Impedance, pp. 51495154, 2006.
[24] H. Xin, Z. Qu, J. Seuss, and A. Maknouninejad, A self-organizing
[1] D.T. Nguyen, L.B. Le, Risk-Constrained Profit Maximization for Micro- strategy for power flow control of photovoltaic generators in a distribution
grid Aggregators with Demand Response, Smart Grid, IEEE Transactions network, IEEE Trans. Power Syst., vol. 26, no. 3, pp. 14621473, 2011.
on, vol. 6, no. 1, pp. 135146, 2015. [25] F. Rahimi and A. Ipakchi, Demand response as a market resource under
[2] Y.A. Muoz, Optimizacin de recursos energticos en ZNI mediante estrate- the smart grid paradigm, IEEE Trans. Smart Grid, vol. 1, no. 1, pp.
gias de suministro y consumo, Disertacin Ph.D., Dept. Eng. Elect., Univ. 8288, 2010.
Politcnica de Valencia, 2012. [26] U.S.E.I.A, Levelized Cost and Levelized Avoided Cost of New Gen-
[3] T.S. Ustun, C. Ozansoy, and A. Zayegh, Recent developments in micro- eration Resources in the Annual Energy Outlook 2015, U.S. Energy
grids and example cases around the world - A review, Renew. Sustain. Information Administration, 2020 Levelized Costs AEO 2015, June, 2015.
Energy Rev., vol. 15, no. 8, pp. 40304041, 2011. [27] Y.M. Atwa, E.F. El-Saadany, M.M.A. Salama, and R. Seethapathy,
[4] I. Patrao, E. Figueres, G. Garcer, and R. Gonzlez-Medina, Microgrid Optimal Renewable Resources Mix for Distribution System Energy Loss
architectures for low voltage distributed generation, Renew. Sustain. Minimization, Power Systems, IEEE Transactions on, vol. 25, no. 1, pp.
Energy Rev., vol. 43, pp. 415424, 2015. 360370, 2010.
[5] H. Laaksonen, Advanced islanding detection functionality for future [28] H.S.V.S.K. Nunna, and S. Doolla, Demand Response in Smart Distri-
electricity distribution networks, IEEE Trans. Power Deliv., vol. 28, bution System with Multiple Microgrids, Smart Grid, IEEE Transactions
no. 4, pp. 20562064, 2013. on, vol. 3, no. 4, pp. 16411649, 2012.
[6] D. Velasco, C.T. Rodrguez, and J.H. Mora, Algoritmo de deteccin en [29] T. Dragicevic, J. Vasquez, J.M. Guerrero, and D. krlec, Advanced LVDC
modo isla de un inversor, basado en la medicin del segundo armnico, Electrical Power Architectures and Microgrids, IEEE Electrif. Mag., vol.
Rev. Lasallista Investig., vol. 10, no. 1, pp. 6274, 2013. 2, no. March, pp. 5465, 2014.
[7] C.L. Trujillo, D. Velasco, D. Fuente, E.F. Amors, and G.G. Sanfeli, Anlisis [30] S.P. Chowdhury, S.P. Chowdhury, and P. Crossley, Microgrids and
de un mtodo activo para la deteccin de isla basado en la perturbacin de Active Distribution Networks, London, United Kingdom: The Institution
la fase en el pll y comparacin con otros mtodos activos, Rev.Cientfica, of Engineering and Technology, London, United Kingdom, 2009.
no. 12. pp. 113-124, 2010. [31] K. Hongseok and M. Thottan, A Two-Stage Market Model for Microgrid
[8] H.A. Gabbar, A.A. Abdelsalam, Microgrid energy management in grid- Power Transactions via Aggregators, Bell Labs Tech. J., vol. 16, no. 3,
connected and islanding modes based on SVC, Energy Convers. Manag, pp. 101107, 2011.
vol. 86, pp. 964972, 2014. [32] H.S.V.S.K. Nunna and S. Doolla, Energy management in microgrids
[9] R. H. Lasseter, Smart distribution: Coupled microgrids, Proc. IEEE, vol. using demand response and distributed storage - A multiagent approach,
99, no. 6, pp. 10741082, 2011. IEEE Trans. Power Deliv., vol. 28, no. 2, pp. 939947, 2013.
[10] M. Yazdanian, A. Mehrizi-sani, Distributed Control Techniques in Mi- [33] M. Ding and K. Luo, A multi-agent energy coordination control strategy
crogrids, Smart Grid, IEEE Transactions on, vol.5, no.6, pp.2901,2909, in microgrid island mode, Lect. Notes Electr. Eng., vol. 238 LNEE, no.
Nov. 2014. 11511075, pp. 529536, 2014.
[11] D.E. Olivares, A. Mehrizi-Sani, A.H. Etemadi, C.A. Caizares, R. Iravani, [34] A. Parisio and L. Glielmo, Energy efficient microgrid management using
M. Kazerani, A.H. Hajimiragha, O. Gomis-Bellmunt, M. Saeedifard, Model Predictive Control, IEEE Conf. Decis. Control Eur. Control Conf.,
R. Palma-Behnke, G.A. Jimnez-Estvez, and N.D. Hatziargyriou, Trends pp. 54495454, 2011.
in microgrid control, Smart Grid, IEEE Transactions on, vol.5, no.4, [35] W. Shi, X. Xie, C. Chu, and R. Gadh, A Distributed Optimal Energy
pp.1905,1919, 2014 Management Strategy for Microgrids, in IEEE SmartGridComm, pp. 16,
[12] R.G. Wandhare, S. Thale, and V. Agarwal, Design of a photovoltaic 2014.
power conditioning system for hierarchical control of a microgrid, Photo- [36] H. Tazvinga, B. Zhu, and X. Xia, Energy dispatch strategy for a
voltaic Specialist Conference (PVSC), 2014 IEEE 40th, pp. 31443149, photovoltaic wind diesel battery hybrid power system, Sol. Energy,
2014. vol. 108, pp. 412420, 2014.
[13] S. Thale, R.G. Wandhare, and V. Agarwal, A Novel Reconfigurable [37] C. Bordons, F. Garca-Torres, and L. Valverde, Gestin ptima de la Energa
Microgrid Architecture with Renewable Energy Sources and Storage, In- en Microrredes con Generacin Renovable, Rev. Iberoam. Automtica e
dustry Applications, IEEE Transactions on , vol.51, no.2, pp.1805,1816, Informtica Ind. RIAI, vol. 12, no. 2, pp. 117132, 2015.
2015. [38] J. Ma, S. Member, F. Yang, Z. Li, and S. J. Qin, A Renewable
[14] E. Mojica-Nava, C.A. Macana, and N. Quijano, Dynamic Population Energy Integration Application in a MicroGrid Based on Model Predictive
Games for Optimal Dispatch on Hierarchical Microgrid Control, Syst. Control, Power Energy Soc. Gen. Meet. 2012 IEEE, pp. 16, 2012.
Man, Cybern. Syst. IEEE Trans., vol. 44, no. 3, pp. 306317, 2014. [39] F. vila, D. Sez, G. Jimnez-estvez, L. Reyes, and A. Nez, Fuzzy demand
[15] F. Katiraei, R. Iravani, N. Hatziargyriou, and A. Dimeas, Microgrids forecasting in a predictive control strategy for a renewable-energy based
management, IEEE Power Energy Mag., vol. 6, no. june, pp. 5465, microgrid, in 2013 European Control Conference (ECC), no. 1, pp.
2008. 20202025, 2013.

513
 2015 IEEE PES Innovative Smart Grid Technologies Latin America (ISGT LATAM)

[40] [40] M. Paulescu, O. Mares, E. Paulescu, N. Stefu, A. Pacurar, D. Cali-

noiu, P. Gravila, N. Pop, and R. Boata, Nowcasting solar irradiance using
the sunshine number, Energy Convers. Manag., vol. 79, pp. 690697, 2014.
[41] M. Lipperheide, J.L. Bosch, and J. Kleissl, SE-D-14-00153 Review 15
march - Embedded nowcasting method using cloud speed persistence for
a photovoltaic power plant, Sol. Energy, vol. 112, no. SE-D-1400153, p.
16, 2014.
[42] S.R. West, D. Rowe, S. Sayeef, and A. Berry, Short-term irradiance
forecasting using skycams: Motivation and development, Sol. Energy, vol.
110, pp. 188207, 2014.
[43] H. Breitkreuz, M. Schroedter-Homscheidt, T. Holzer-Popp, and S. Dech,
Short-range direct and diffuse irradiance forecasts for solar energy ap-
plications based on aerosol chemical transport and numerical weather
modeling, J. Appl. Meteorol. Climatol., vol. 48, no. 9, pp. 17661779,
2009.
[44] M. Lydia and S.S. Kumar, A comprehensive overview on wind power
forecasting, IPEC 2010 Conf. Proc., vol. Singapore, pp. 268273, 2010.
[45] T. Brandt, N. DeForest, M. Stadler, and D. Neumann, Power Systems
2.0 Designing an Energy Information System for Microgrid Operation,
in Proceedings of 2014 International Conference on Information Systems
(ICIS 2014), pp. 118, 2014,
[46] Y. Ren, P.N. Suganthan, and N. Srikanth, Ensemble methods for wind
and solar power forecastingA state-of-the-art review, Renew. Sustain.
Energy Rev., vol. 50, pp. 8291, 2015.
[47] C.A. Ramrez Escobar, Los Precios Del Mercado Mayorista de Electrici-
dad como Expresin de la Participacin Activa de la Demanda: Aplicacin de
la Economa Experimental, Disertacin PhD, Dto. Ing. Electr., Universidad
Politcnica de Valencia, 2008.
[48] L. Tao,C. Schwaegerl, S. Narayanan, and J.H. Zhang, From laboratory
Microgrid to real markets Challenges and opportunities, eighth Int. Conf.
Power Electron. - ECCE Asia ”Green World with Power Electron., ICPE
2011-ECCE Asia, pp. 264271, 2011.

514