Domestic Load Management With Coordinated Photovoltaics, Battery Storage and Electric Vehicle Operation

Coordinated power demand management at residential or domestic levels allows energy participants to efficiently manage load profiles, increase energy efficiency and reduce operational cost. In this paper, a hierarchical coordination framework to optimally manage domestic load using photovoltaic (PV) units, battery-energy-storage-systems (BESs) and electric vehicles (EVs) is presented. The bidirectional power flow of EV with vehicle to grid (V2G) operation manages real-time domestic load profile and takes appropriate coordinated action using its controller when necessary. The proposed system has been applied to a real power distribution network and tested with real load patterns and load dynamics. This also includes various test scenarios and prosumer’s preferences e.g., with or without EVs, number of EV owners, number of households, and prosumer’s daily activities. This is a combined hybrid system for hierarchical coordination that consists of PV units, BES systems and EVs. The system performance was analyzed with different commercial EV types with charging/ discharging constraints and the result shows that the domestic load demand on the distribution grid during the peak period has been reduced significantly. In the end, this proposed system’s performance was compared with the prediction-based test techniques and the financial benefits were estimated.

The net power gained by EV.    Base load. C BES E Cost of the energy supplied by the battery storage system. µ DoD Depth of discharge.

I. INTRODUCTION
Traditionally, power grids are centrally managed and supply electricity to consumers through a network of long-distance transmission and distribution lines. Conventional power systems nowadays see a significant paradigm shift towards an automated and distributed trend of energy management. The huge influx of small-scale and aggregated renewable energy resources (e.g., solar, wind) at the domestic or distribution level creates multiple power management challenges to the utilities [1], [2], [3]. This is because most of these renewable sources are intermittent in nature, and so maintaining the power quality according to standard voltage and frequency requirements is a significant challenge [2], [3].
As a new type of dynamic load, electric vehicles (EVs) are also now being integrated into the power networks. Unlike standalone battery energy storage (BES) systems, the mobility of EVs makes their in-built energy storage capability more dynamic than the fixed installation storages [4], [5], [6]. The bidirectional power flow from an EV, in residential areas or aggregated EVs in a parking lot through grid-to-vehicle (G2V) and vehicle-to-grid (V2G) operation, has the potential to provide ancillary support and various loads to the power grids [6], [7], [8]. Moreover, the adverse effects due to the intermittent behavior of renewables and vehicle's uncertainty related to their availability can be lessened with the combination of EVs and fixed energy storages [2], [4]. To reduce the energy usage cost, these energy storages perform charging operations during off-peak periods (i.e., the duration of low energy cost), while during peak-load periods (i.e., the duration of high electricity demand and price), their discharging operation potentially assists in reducing the peak-load at domestic sides [4], [8]. Any load profile is likely to have multiple increasing peaks over time; therefore, if these peak loads are not coordinately or regularly managed on domestic sides, these will cause adverse effects to nearby power network components (e.g., distribution transformer, power cables, and protection circuitry) [4], [5], [6], [7].
In this context, a coordinated or regulated power demand management is essential to minimize the peakload effects and maximize the utilization of distributed resources [9], [10], [11]. Additionally, it reduces the energy costs for customers during peak-load periods and incentivizes customers when there is less demand [8], [11]. With coordinated management, sometimes various forecasting techniques are implemented for the purpose of scheduling as well as optimizing the energy resources [2], [12]. Thus, it has the potential benefits of improving power quality (i.e., increasing PV voltage and reducing flicker voltage), stability management (i.e., voltage and frequency) and power regulation (balancing power flow) [9], [10], [12]. For domestic load management, till now, the majority of previous work is either considered rooftop PV units with different household battery sizes or mere integration of EVs. Domestic BESs used in a small-scaled microgrid provide various ancillary support considering managing load, frequency or voltage [4], [13]. The peak-load management with EVs in commercial systems is elaborately reported in [11]. The impacts of integrating large-scale battery-storage system for managing increasing demand are also examined in [9] and [13]. In [14], using a decision-tree-based algorithm, a combination of PV, EV, and different battery systems are used at domestic levels to manage the peak load; however, the contribution from each energy source at any time interval and economic aspects are found to be inadequate. The particle swarm optimisationbased energy management system has been applied among the smart home, EVs and PV farm without the consideration of BESs [15]. While, the reduction of peak load using scheduling has been done in [16], without discussing the economic benefit. A hybrid approach for energy demand forecast using autoregressive integrated moving average model has been studied and reported in [17], [18], [19], and [20].
Moreover, the majority of previously developed systems were applied only for a short duration; but the effectiveness of these systems in reducing the peak load considering the real test case scenario for a longer duration is still unclear [8], [21]. For example, in [22], authors developed a load curve manually and scheduled the load with respect to energy rates and PV generations but did not consider any real scenario. Therefore, it has a huge interest to examine the effectiveness of a hybrid system consisting of an EV, PV, and battery storage on real test environments over longer periods. In other words, each resource's contribution at the domestic level with respect to time intervals and load dynamics is a subject yet to be explored. In this paper, a coordinated system consisting of PV units, BES systems and EVs is considered for managing the domestic peak load demand. A hierarchical coordination framework is developed and tested on a real environment of distribution networks in Australia. Three types of commercial EVs with small (Mitshubishi-i-MiEV), medium (Nissan Leaf) and large (Tesla-3) battery sizes are considered for load management. An improved scheduling algorithm is also developed for their charging and discharging operation, while the intended future extension of this project will include the performance comparison of various optimization algorithms applications in the proposed microgrid. For analyzing the economic aspects, real-time load pricing from the Australian energy market is used to study the financial benefits of the proposed management framework. To sum up, the main contributions of this paper are as follows: • Develop an improved domestic energy management framework to manage the domestic load profile and utilize resources in a more efficient and economical way.
• Identify a coordinated approach considering each dynamic source constraint and load contribution for managing distributed resources and peak demand.
• Investigate the peak load reduction process under real test scenarios of uncertainties and seasonal conditions by considering dynamic and environment-dependent sources and load profiles for a longer duration.
• The acceptance of the proposed system is verified with various test case conditions, such as EV types, source combinations, number of EV owners, number of households, prosumer's daily activities etc.
• Compare the performance of the developed system with moving average (MA) and autoregressive integrated moving average (ARIMA) prediction systems.
• Study the financial aspects of the designed framework.

II. SYSTEM OVERVIEW
The schematic diagram of a small-scaled microgrid is shown in Fig. 1. The microgrid model comprises residential households (or loads) and energy sources (a PV panel, battery storage, and an electric vehicle). Depending on the prosumer and the number of households, EVs and BES's number will be more than one. All energy sources are connected to a common AC bus through power electronic converters. Both the BES and EV have bidirectional power flow capabilities. A central controller is attached to the residential places which regularly monitors the load pattern, PV power generation, load demand, BES, EV's state of charge (SOC), and chargingdischarging condition. Depending on load requirements, conditions, and resources (i.e., PV, battery and EV) availability, the controller will make the charging or discharging operation of both BES and EV either from the PV or from the AC bus, respectively.

A. ELECTRIC VEHICLE SPECIFICATIONS
There are three commercial V2G capable electric vehicles with different battery sizes (i.e., small, medium, and large) considered for load management. These are Mitsubishi-i-MiEV, Nissan Leaf and Tesla Model 3. Each car belongs to a single household. The total energy consumed or served by the EV depends on the user profile of the EV owner [23], [24]. To comply with the prosumer's habit in real life, three different charging profiles (for people going to work during the daytime, people staying at home, and people working on night shifts) are considered. For the daytime working profile, the vehicle owner leaves home at 9.00 am and returns at 4.00 pm. There are no charging facilities available at the workplace parking area. To avoid any undesirable situation, the EV battery will never fall below the minimum state of charge (SOC) level. Here, all assumptions are taken into account for simulation, and analysis purposes only and these can be modified based on a more available realistic scenario. For example: in [25], the authors build a hybrid (power grid and communication) model and considered the EV's status during the transit period also, which brings a more realistic view of the practical scenario. In this work, for simplicity, the instantaneous SOC of the EVs during the transit period has not been considered. However, all key parameters related to the EV specifications are given in Table 1. The station, process and bay level are connected to each other by two buses-the process bus and station bus [26].

B. BES AND PV's SPECIFICATION
Each house has its own PV panel and BES system. For simulation purposes, it is assumed that the energy generation from each house (i.e., either PV panel or BES) is similar. The detailed specifications of the PV panel and BES are listed in Table 2.

III. MATHEMATICAL MODEL FOR ENERGY MANAGEMENT
Overall energy management of the whole microgrid system is performed following an optimization algorithm implemented in a central controller. The detailed architecture of the whole microgrid system is shown in Fig. 2. The central controller is connected to various smart meters and converters (Conv-1, 2, 3, 4). Conv-4 (dc-dc) and Conv-2 (dc-ac) connect the PV panel to the AC bus, while Conv-3 (dc-dc) connects the BES with PV via a DC bus when there is a surplus PV generation during daytime. On the other hand, both the battery and EV are connected to the AC bus via bidirectional power flow functionality type (i.e., DC-AC and AC-DC) converters.    The power flow (i.e., inward or outward current) of these converters entirely depends on the residential load conditions and local energy generation. On the other hand, the controller regularly reads the domestic load demand, power generation, availability of resources and the real-time electricity cost from the utility to check the peak load managed energy price steps. The implemented optimization technique follows a set of mathematical programming problems considering technical or system constraints of PV units, BESs, and EVs integrated into the grid. The overall aim is to reduce total systems costs and ensure efficient use of energy resources. The stepby-step mathematical models for each network component are described in the following sub-sections.

A. EV WITH ENERGY TRANSACTION
An EV participates in energy transactions through the charging or discharging process of its in-built battery storage. The charged or discharged amount is limited by two boundary points: (a) minimum state of charge level (ψ EV min ) and (b) maximum state of charge level ψ EV max . Depending on load conditions, the EV will go only to V2G operation if it exceeds the certain threshold limit (ψ EV max ) usually set by the manufacturers. Moreover, between the minimum SOC level ψ EV min and maximum charging limit ψ EV max , the charging power relates to the state of charge level by a linear piecewise function and can be expressed as of (1), as shown at the bottom of the page, [8], [14]: In this paper, any contribution to the grid (i.e., power injected into the grid) is shown with +'ve superscript, while any power taken from the grid is labelled with 've superscript.
Here, for the V2G operation, the maximum available power from an EV during the discharging process can be expressed as follows: Let us assume, be a set of periods for which an EV of a microgrid system is plugged for either G2V or V2G operation, where, X means the total number of EVs of the concerned area. The limiting constraints of charging or discharging power can be written as follows: Here, the binary variable EV t,χ represents the charging or discharging state/mode of an EV. The value EV t,χ varies for the following conditions: On the other hand, t,χ ≥ 0 denote the charged power to the battery and the discharged power from the battery to the grid respectively.
When an EV completes the journey for the purpose of work and returns home, the energy required by the vehicle is ℓ EV + χ for the period of ∀t ∈ T \ξ χ . Then, the net energy p EV t,χ at any time interval τ can be written as follows: Here, η 1 denotes the efficiency of converter 1. The net power gained p EV t,χ by an EV in relation to the SoC level at an interval τ can be calculated from the following set of equations.
Here, χ represents the battery capacity of an EV and ψ EV I ,χ denotes the instantaneous SOC of EV. Note that each house will maintain a maximum of one vehicle in a plugged-in state at each period t ∈ T . Therefore, the number of EVs plugged into a microgrid system will be equal to or lower than the number of households κ. This can be expressed by following constraints.
and plug t,χ ≡ ≥ 1; ∀t ∈ ξ χ , ∀χ ∈ X 0; ∀t ∈ T \ξ χ , ∀χ ∈ X For plug t,χ ≥ 1, the microgrid must be either receiving or feeding energy to the EVs; however, any plugged-in EV will not participate in both energy transactions at the same time [27]. Therefore, the total electric power P EV tot derived from all EVs (∀χ ∈ X ) in a microgrid system at any test period (∀t ∈ ξ χ ) can be written as: The domestic load curve is always a function of power (p) and time (t). Note that the total domestic load demand at the common AC bus must be equal to or greater than the total consumers' power demand connected to that bus. Since the phase voltage at that common bus does not vary significantly with the load change, so only the load current varies depending on the consumers' power demand.

B. MICROGRID POWER DEMAND
In a microgrid system, the domestic power demand ℜ p t and the consumers' desired base load demand ℜ p t are always a function of independent variables time (t) and power (p), and are expressed as below [28]: Domestic BES and EVs always read base load (off-peak) and peak-load periods for their charging and discharging operation. The load periods (peak/ off-peak), in sub-second timescale, are usually sensed by the controller. These may occur several times within a finite duration regardless of whether it is a day or month. Let's assume the peak load starts at τ p s and ends at τ p e , while the off-peak load initiates at τ b s and stops at τ b e . If the peak load and off-peak load occurrence frequency are designated as m and n, respectively, then the total time period T can be expressed as the summation of both peak-load (τ p ) and base load (τ b ) periods and is given by The difference between the domestic power demand (ℜ p t ) and consumers' baseload demand (ℜ p t ) determines the offpeak and peak-load periods. The required power (P req t ∈ ℜ p t ) demand, whether peak or off-peak, at t th instant (t ∈ τ p , τ b ) can be written as follows: The peak load and baseload periods are determined by the following equation, P req t ≡ ≥ 0; peak-load demand condition < 0; off-peak demand condition (17)

C. PEAK-LOAD CONDITION
For P req t ≥ 0 at t th instant, the power difference is supplied by the available energy resources to balance the load demand. In this case, the energy sources (i.e., PVs, BESs, EVs) will provide the required power to the common AC bus. Let P BES + tot and P PV + tot be the maximum power available from the BESs and PV units, respectively. Hence, the power balance equation of any microgrid system can be written as follows: If P EV tot = 0; the EVs are not in a plugged-in state and not available to act as either a source or sink, the mismatched power P req t − P PV + tot , P EV tot , P BES + tot is supplied from the AC bus. Assume the lower SOC boundary and the instantaneous SOC for a BES are ψ BES min and ψ BES t , respectively. Therefore, the maximum power received from a battery unit installed in a household ψ BES t ≥ ψ BES min is given by the following equations: Here, BES is the capacity of the battery storage unit. η 2 and η 3 denote the efficiencies of converter 2 and converter 3 respectively, P PV + κ is the average PV power generation per household and η 4 denotes the efficiency of converter 4. VOLUME 11, 2023

D. OFF-PEAK CONDITION
For off-peak periods (P req t < 0), the available grid power P grid t can be used to charge the energy storages (i.e., EVs and BESs), so under this condition, the power-balance equation can be written as follows: In this case, the amount of power required to charge a BES varies according to the difference between the maximum charging limit and SoC status. So, the maximum charging power requested by a BES (p Here, ψ BES max represents a maximum charging limit of battery storage. On the other hand, EVs will charge up to the requested power according to (1).

E. POWER SOURCES, LOADS AND POLARITY
During the peak load and off-peak load periods, depending on the charging and discharging process, the number of energy sources and loads vary in order to balance the power-demand equation. At any time instant (t ∈ τ p ∪ τ b ∈ T ), the total supply sources (P S tot ) and loads (P L tot ) connected to the main AC bus can be written as follows [28].
For P req t < 0(off-peak), EVs (26) and for P req t ≥ 0(peak-load), is defined by equation (27). The polarity of power flow from any energy source indicates whether it will act as a source or sink. In other words, any positive polarity in the simulation results represents the action as a source; conversely, negative polarity indicates the load operation. Moreover, the power flow polarity identifies the charging and discharging mode of an EV and battery.

F. ALGORITHM OF POWER DEMAND MANAGEMENT
The algorithm proposed to manage the domestic load is shown in Table 3 followed by two flowcharts as in Fig. 10 and Fig. 11. It includes the coordinated operation of EV, BES and PV for shaving the power demand of a real-time load curve.

IV. CASE STUDIES ON A REAL NETWORK
The total working methodology of this paper has been depicted in Fig.3. While, Fig. 4 shows the proposed control framework which has been applied to a real power distribution network located in the Nelson Bay area of New South Wales (NSW), Australia. All relevant information on network configuration, load pattern, demand response and available resources were obtained from a trial 'Smart Grid, Smart City (SGSC) project funded by the Australian Government in collaboration with Energy Australia, Ausgrid, and other consortium partners like Nelson Bay City Council, Sydney Water, GE Energy Australia etc. [26], [29]. The majority of loads in this area are more likely residential, rural and small industries and are serviced by two 33-kV substations located in Tomaree and Nelson Bay, NSW.
The efficacy of the proposed system and control framework for managing the domestic loads was tested on three neighbouring households (H1, H2 and H3). These customers are powered through a long feeder from Nelson Bay substation and experience seasonal load changes. Climate data were also used here to evaluate the average PV unit's generation and weather dependent load pattern. The average load curves of H1, H2 and H3 in three consecutive days are illustrated in Fig. 5 and the corresponding analysis with several test case scenarios is described in the following subsections.
The average PV generation at each house is 3.3 kW and each house has placed 40-kWh battery storage. In this paper, the available baseload (or average) demand is considered as 9 kW from 12.00 pm-11.59 pm and 6.8 kW from 12.00 am-11.59 am. Therefore, the peak load condition is more likely to be observed from 7 am to 9 am in the first half (because of lower PV generation) and from 6 pm to 10 pm in the second half of any test period day (because of higher domestic load demand). These assumptions are realistic, completely complying with the actual load demand pattern given by the Australian Energy Market Operator (AEMO) [30]. Initially, the PV power is not available, and the battery storage is charged up to the maximum level. To enhance the battery lifetime by reducing the depth of discharge (DoD), the BES charging/ discharging limit is maintained within 40 ∼ 95% of the SOC level [31]. In this research, 60% of DoD has been considered.

A. LOAD MANAGEMENT WITH PV AND BES
The load demand management of three households with controlled PV and battery storage operation is shown in Fig. 6. The PV power generation is based on the irradiation curve found in [32]. As can be observed, with the controlled operation, during the off-peak period the households receive a large proportion of active power from either battery storage (especially, at night-time) or PV units (especially, during daytime). If any surplus energy generation happens, the microgrid system returns additional power to the grid or charges the BES. In Fig. 6, this can be identified as the negative polarity of the grid and battery power. On the other hand, the consumption of grid power reaches a maximum during the peak period because the battery storage is solely not capable to supply any additional power requirement to balance the load demand. It happens two times per day, first in the morning and second happens after the evening when PV power is absent.

B. LOAD MANAGEMENT WITH PV, BESs AND EV TYPES
In this case study, the combined response for managing domestic loads using the PVs, battery storages and a commercial EV from three different types: Mitsubishi, Nissan Leaf and Tesla-3 is discussed. As described earlier, these commercial EVs have their own in-built small, medium and larger battery and their coordinated response are shown in Fig. 7, Fig. 8 and Fig. 9 respectively. From Fig. 7, it is evident that during the peak period after battery storage operation, if any additional power is required, the controlled system, instead of taking power from the grid, mitigates the required energy demand with the V2G operation of the Mitsubishi. Unlike Nissan Leaf and Tesla-3 vehicles, due to smaller battery sizes, the Mitsubishi performs more charging and discharging within 24 hours. It is because it can reach its highest and lowest SOC limit faster than the other two types of cars. In fact, for its lower charge capacity, it needs a charging operation after each V2G operation. On the other hand, during the off-peak period occurs, if the EV is in the pluggedin state, then it will be likely to be charged from available resources. For EVs, here the constant voltage constant current charging topology has been considered; hence the charging power is constant.
In Fig. 7, Fig. 8 and Fig. 9, during 10.00 am -5.00 pm, PV generation is maximum, so in this period, after managing the domestic load and charging the BESs, the excess PV energy is directly transferred to the AC bus through the converters 2 and 4. If an EV is available at home after coming from work (4.00-7.00 pm) and the PV and BES power are not sufficient to charge the EV and manage the load demand, then the shortfall is drawn from the common AC bus.
As expected, and observed in Fig. 9, due to the larger battery size, the Tesla-3 model manages more load demand and thereby, provides comparatively a higher load factor during the peak period than that of other EV types. The effectiveness of the proposed system is also tested with the load conditions of two households and found similar results of managing the load demand regardless of conditions whether it is peak or off -peak period.

C. THREE HOUSEHOLDS HAVING EVs
Instead of having a single EV, the proposed system is considered with three EVs, where each house has just one EV type and the combined response is shown in Fig. 12. It is found that the developed system is independent to load pattern and network component's constraints. However, the power consumption from the AC bus, in this case, is found to be higher and less V2G duration is observed due to the higher aggregated battery power consumed by all EVs. The charging curves are not constant for the whole charging period, because, in the graph, the combined charging power is shown. Where Tesla-3 is charging for the longest period for its highest battery capacity.

V. COST-BENEFIT ANALYSIS
This section aims to demonstrate the cost-benefit analysis of the domestic load management system. Let's assume the unit cost of the energy supplied by the battery storage system is

C BES
EĈ BES E . It is calculated by dividing the total installation and maintenance cost of the BESs by the total energy generated from BESs (W BES A ), and can be written as follows [11]: where, W BES A can be found by multiplying the instantaneous discharged power, the number of the charge-discharge cycle with the discharging time length and the operating duration in a year (in days). Mathematically, the W BES A can be written as: The per-year battery cost (Ĉ B ) can be formulated by adding together the per unit battery cell (C b,pu ) cost, the battery management cost (C m,pu ) and the cost of power equilibrium (C eq ). It can be expressed as follows.
The total battery converter cost P BES C can be calculated by multiplying the per unit battery converter cost with the BES capacity. If the installation, maintenance, and operation cost VOLUME 11, 2023 for each kWh of battery storage isĈ f ,B , the total installation, maintenance, and operation cost C f ,B can be calculated by the following (31), It is also needed to calculate the salvage value of the storage battery and deduct it from the energy cost of the BESs. If the initial and lost capacity of the battery are Q i and Q loss respectively, then the salvage value of the BESs S C is It should be noted that the battery capacity can be lost due to cycle or calendar fading or ageing. Battery ageing is governed by various charging/discharging and chemical factors and therefore, an accurate prediction of the battery ageing model is quite difficult. A semi-empirical approach is proposed in [33], where the Q loss has been related to the discharging current rate (I d ), and charge output (A H ) as follows: Here, β is a pre-exponent factor which is inversely related with I d . The complete charging/discharging impact (or degradation) on the battery life L d by considering the temperature  impacts µ T and depth of discharge (DoD) µ DoD is given as: Here, L rat is the rated battery life cycle. Based on the AEMO electricity price during peak and off-peak periods in 2020 [30], the cost-benefit analysis of the proposed system with any of the Mitshubishi, Nissan Leaf and Tesla EVs is illustrated in Fig. 13. It can be observed that the coordinated operation of EV shows much better performance in terms of energy cost than that of the unregulated management. Note that the electricity price data here are based on the summer season of NSW in Australia and fed directly into the controller to calculate the total energy cost. It is shown that, the energy price has been reduced to 38.18% due to the regulated use of Mitshubishi in the summer season, which is 6.5% more than the article [15]. The benefit will grow more for the longer use of the proposed energy management system.

VI. COMPARISON WITH OTHER MODELS
With the controlled operation of PV, BESs and a Nissan Leaf EV, the proposed system is compared with different prediction-based controlled techniques, such as moving average (MA) and autoregressive integrated moving average system (ARIMA). Autoregressive model predicts depending on the previous error, so it does not change instantaneously. Moving average (MA) is a better process as it depends on the past value of prediction and had been used for years for load forecasting [34]. Autoregressive MA (ARMA) model predicts data using both error and predicted past value. The autoregressive integrated MA (ARIMA) is a more advanced tool compared to ARMA. The ARIMA Model is especially appealing since it uses logical and well-organized processes to create models utilizing the autocorrelation and partial autocorrelation functions [35]. During load forecasting, the  relatively weak response of the ARIMA model to sudden disturbances makes it more preferable to implement [36]. In this research, a simple model MA and an advanced model ARIMA have been considered. The comparison is depicted in Fig. 14 and it confirms that the proposed control strategy shows similarly improved performance in reducing the peak load demand with other prediction techniques.

VII. CONCLUSION
This paper presents an improved framework of managing domestic load demand with the coordinated operation of photovoltaics (PVs), battery energy storages (BESs) and electric vehicles (EVs). Three commercial EVs with small (Mitshubishi i-MiEV), medium (Nissan Leaf) and large (Tesla) battery sizes with respect to the domestic load pattern in a real power distribution network were used to examine and validate the designed network performance. These studies showed that, under realistic circumstances, the developed framework can significantly reduce the peak load demand on the AC main grid. Moreover, an EV with the larger battery capacity can manage more peak domestic loads comparing to other EV types, thus suggesting an improved utilization of electricity infrastructure.
The findings were further supported by the cost-benefit analysis using the real energy price given by a local energy distributor. Under these realistic load patterns, the developed framework was also compared with prediction-based techniques and similar improved performances in managing domestic energy demand were found. These results will be useful for both efficient and economic utilization of distributed energy resources and their coordinated management.