0% found this document useful (0 votes)
15 views

IJEEE - Microgrid BSS Scheduling Using Teaching Learning Based Optimization Algorithm

Energy storage serves as a crucial hub for the entire grid, supplementing resources such as wind, solar, and hydropower, as well as nuclear and fossil fuels, demand side resources, and system efficiency assets. It can function as a generation, transmission, or distribution asset — all in one unit. Storage is, in the end, an enabling technology. It has the potential to save consumers money while also improving reliability and resilience, integrating power sources, and reducing environmental impac

Uploaded by

iaset123
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
15 views

IJEEE - Microgrid BSS Scheduling Using Teaching Learning Based Optimization Algorithm

Energy storage serves as a crucial hub for the entire grid, supplementing resources such as wind, solar, and hydropower, as well as nuclear and fossil fuels, demand side resources, and system efficiency assets. It can function as a generation, transmission, or distribution asset — all in one unit. Storage is, in the end, an enabling technology. It has the potential to save consumers money while also improving reliability and resilience, integrating power sources, and reducing environmental impac

Uploaded by

iaset123
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 11

International Journal of Electrical and

Electronics Engineering (IJEEE)


ISSN (P): 2278–9944; ISSN (E): 2278–9952
Vol. 10, Issue 2, Jul–Dec 2021; 5–14
© IASET

MICROGRID BSS SCHEDULING USING TEACHING LEARNING BASED


OPTIMIZATION ALGORITHM

Yendru Sravani, Dr. M. Veera Kumrai & N. Rama Narayana


Research Scholar, Sir CR Reddy College of Engineering, Andhra Pradesh, India

ABSTRACT

Energy storage serves as a crucial hub for the entire grid, supplementing resources such as wind, solar, and hydropower, as well
as nuclear and fossil fuels, demand side resources, and system efficiency assets. It can function as a generation, transmission, or
distribution asset — all in one unit. Storage is, in the end, an enabling technology. It has the potential to save consumers money
while also improving reliability and resilience, integrating power sources, and reducing environmental impacts.

Battery storage system design is now important for microgrids to prepare a day-ahead schedule for steady
operation. This article discusses the scheduling of BSS, which helps to reduce the average cost imposed on microgrid
consumers in the context of dynamic pricing. For minimizing, a cost function is created and subjected to optimization
based on the restrictions. The search space magnification is 50*(DC– DD + 1), where DC and DD are the maximum charge
and discharge depths in an hour in percentage for a specific BSS, respectively. The programming is done by combining
daily load, generated energy, and grid price forecasts with a microgrid size as specified in the article and implementing
Teaching Learning Based Optimization (TLBO) for achieving an average cost reduction when compared to Net Power
Based Algorithm and Particle Swarm Optimization for a planned BSS.

KEYWORDS: Important for Microgrids, Implementing Teaching Learning Based Optimization (TLBO)

Article History
Received: 27 Oct 2021 | Revised: 28 Oct 2021 | Accepted: 08 Nov 2021

INTRODUCTION

An algorithmic approach for the Optimization of Day-Ahead Energy Storage System Scheduling in Microgrid problem by
using Genetic Algorithm and Particle Swarm Optimization. Both GA and PSO fared well compared to the Net Power
Based Algorithm[1].

A grid is a system for transferring electricity from sources to consumers. With the aid of contemporary smart
technologies, traditional power grids, which can only transport energy in one direction, from producer to consumer, are
rapidly developing into a two-way power flow system [2]. A smart grid is a bi-directional energy transmission and data
communication network that is smart and dispersed [3]. The smart grid is expected to develop as a result of the plug-and-
play integration of smart microgrids [4].

A microgrid is like a mini version of a power grid [5]. According to the US Department of Energy, a microgrid is
a collection of interconnected loads and distributed energy resources that operate as a single regulated body in respect to

www.iaset.us [email protected]
6 Yendru Sravani, Dr. M. Veera Kumrai & N. Rama Narayana

the grid within correctly established electrical limitations [6].

When technical or economic conditions warrant it, a microgrid is a distributed group of energy resources and
loads that is mainly associated to and synchronized with the conventional broad range synchronous grid (macro-grid), but
can unplug from the interconnected grid and execute autonomously in island mode. By transitioning between island and
connected modes, microgrids improve supply security within the microgrid unit and can offer backup energy. An off-grid
microgrid, also known as an autonomous, stand-alone, or solitary microgrid, is another use.

Battery storage systems (BSSs) play an important and diversified function in microgrids. A voltage source is
required to synchronize solar PV and other renewable distributed generation (DG) systems. Customarily, a backup
generator has been used for this. An BSS, on the other hand, provides an equivalent energy supply without the pollution
that a diesel generator produces.

BSSs are commonly used to store and utilize electricity generated at various periods. A important concern is that a
BSS in a microgrid deteriorates with time due to frequent charging/discharging cycles [7].

To prolong the length of life of a BSS, a microgrid may undertake the following actions:

Optimal SoC

As per traditional power storage standards, Li-ion batteries are most useful between 10-20% and 80-90% State of Charge
(SoC). Excessive or insufficient use of these restrictions may decrease the battery's life cycle and reduce energy
production.

Depth of Charging / Discharging Limit

The Depth of Charge/Discharge is a proportion of the battery's total capability that reflects the amount of electricity that is
cycled in and out of the battery in a particular time interval. Most batteries have a physical limit on how many times they
can charge and discharge in a particular length of time.

However, to improve battery life, it is recommended that you keep well below this number. For example, a
battery's maximum depth of discharge for one hour may be 30 percent. Any discharge level that is higher than this may
harm the battery.

Dynamic energy pricing is developing traction as a way to promote peak shaving and load levelling by charging
premium rates amid periods of high demand, incentivizing consumers to reduce their consumption and changing the load
curve. Lowering peak demand and levelling demand profiles benefits producers since it decreases overall plant and capital
costs [3]. TOU pricing is a kind of dynamic pricing in which the cost of an electrical unit is influenced by the time of day.

Consumers are likely to devise ingenious ways to take advantage of changing tariffs [8]. One such technique is to
use the BSS to store electricity while the tariff is low and then use the energy when the tariff is higher. Efficient BSS
schedule, or choosing when the BSS charges or discharges, is required for this. Moving particular loads to a various time
points during the day, typically when the grid cost is lowest, is another way to capitalize from dynamic pricing. Load
shifting is a demand response strategy that has been explored in numerous papers [9]-[11].This, like the preceding
approach, is a tough optimization problem. Consider a typical microgrid model that is coupled to a classic one-way-power
grid or classic grid. This is connected to a BSS for the purpose of studying the model.

Impact Factor (JCC): 7.0125 NAAS Rating 2.96


Microgrid BSS Scheduling Using Teaching Learning Based Optimization Algorithm 7

In this article, the first problem will be acknowledged: optimizing the BSS schedule in a microgrid with dynamic
pricing in order to minimize the overall cost charged by consumers. We utilized TLBO [12] to optimize a day-ahead BSS
schedule in a microgrid connected to a conventional one-way-power-flow grid with dynamic pricing. A day-ahead hourly
load, generated electricity, and grid price forecast are required for this optimization problem. Despite the reality that load
and generation energy forecasting is a difficult problem [14], numerous studies have given a variety of strategies for
predicting with reasonable accuracy [15]-[18].

Problem Formulation

An objective function is developed and bounds are provided in this work. We use greedy selection to discover the most
optimal solution to our optimization issue.

Consider a micro-grid comprised of a diverse array of solar panels, wind turbines, and (if necessary) additional
energy generators. The micro-grid also has an unified BSS with capacity CBSS (in kWh). The microgrid is linked to a
normal one-way-power-flow grid (called Utility). The microgrid is subjected to hourly dynamic power pricing by the
traditional grid (also known as the main grid). The load in a micro-grid is the sum of all the loads in the micro-grid. In
order to decrease the cost paid by micro-grid customers, we optimize the BSS charge/discharge schedule for a single day
divided into distinct intervals of time.

Let Xt indicate the microgrid's cumulative load (in kWh), Gt represent the non-conventional energy produced by
the microgrid (in kWh), and Ah denote the unit price for power generation pulled from the power grid (in cents/kWh) even
during hourly interval h where 1 ≤t≤24.

Let DC be the greatest depth of charge achieved by the BSS in one interval. Set DD to the maximum depth of
discharge in the same way. Then we construct a battery schedule vector as follows. A battery planned vector is a 24-
dimensional real-valued vector [S1, S2,...,S24] that represents the day-ahead planning of a microgrid battery. Each St
indicates the depth of charging/discharging in the hour period h and must adhere to the restrictions.

( )
max −1 ∗ ∗ 100, 𝐷 ,𝐷 𝑋 > 𝐺  
𝑆 ∈ (1)
[0, 𝐷 ]𝑋 ≤ 𝐺

10 ≤ ∑ 𝑆 ≤ 90 ∀ 𝑛 ∈ [1,24] (2)

The limitation in Equation 1 ensures that the cell does not discharge much beyond |DD| percent and does not
charge much more than |DC| percent in an hour. The types of batteries used determine the DD and DC values. It also protects
the battery from being discharged if the quantity of power generated is more than the load required. The limitation in
Equation 2 ensures that the battery's SoC does not fall below 10% or climb over 90% to maintain an optimal state of
charge. To determine the costs associated with a battery scheduling vector, we first create the power consumption vector.

Each Ect in an energy consumption vector, which is a 24-dimensional valued vector [Ec1, Ec2,........,Ec24], represents
the amount of energy taken by the microgrid from the grid system (in kWh). Eh comes from the following sources:

𝐸 = 𝑚𝑎𝑥(𝑋 − 𝐺 + (0.01 ∗ 𝐶 ∗ 𝑆 ), 0) (3)

Using Equation 3, the whole input or outflow of energy from the batteries during the time t is approximated by
0.01*CBSS*St (depending on the value of St). When Xt> Gt, Xt - Gt represents the remaining energy required to fully power the
load, or the additional energy left (possibly to charge the battery) after Gt has been used to power the load (when Xt<Gt).

www.iaset.us [email protected]
8 Yendru Sravani, Dr. M. Veera Kumrai & N. Rama Narayana

Because we assume that electricity only flows into the grid and not out of it, the'max' ensures that Ect does not go below zero
(when Xt>Gtand the battery is unable to entirely conserve the surplus power due to the constraints in Equations 1 and 2).

Minimizing the consumer's total cost, TC, which is the total of the product of Ect and the unit electricity prices
during period t, At, over all intervals, becomes the optimization problem.

𝑚𝑖𝑛𝑖𝑚𝑖𝑧𝑒 𝑇𝐶 = ∑ 𝐸 ∗ 𝐴 (4)

The problem's search space has now been specified. It's worth mentioning that we're expecting that charging and
discharging depths advance or decrease in 0.01 increments. In our example, the BSS's SoC can range from 30% to 30%.01
percent, but there is no number in between. In the real world, changes in the SoC of batteries have corresponding minimal
step values. This is referenced to as the charging/discharging minimum depth.

The search space is specified as a set of all 24-dimensional vectors, such as TC = [S1; S2,......, S24], where St is a
real integer in the range [DD, DC], and DD and DC, respectively, represent the highest depth of discharging and charging in
an hour period.

The price of any random search element, such as TC = [S1, S2,......, S24], is defined as Cost(TC).

∑ 𝐸 ∗ 𝐴 𝐼𝑓 𝑈 𝑠𝑎𝑡𝑖𝑠𝑓𝑖𝑒𝑠 𝑐𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑡𝑠 𝑖𝑛 𝑒𝑞. (1) 𝑎𝑛𝑑 (2) 


𝐶𝑜𝑠𝑡(𝑇𝐶) = (5)
∞ 𝑂𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

DESCRIPTION OF TLBO ALGORITHM

We discovered that all evolutionary and swarm intelligence-based optimization approaches are probabilistic, with similar
controlling parameters such as population count, number of generations, elite size, and so on,[12]. In addition to these
common elements, each algorithm has its own set of controlling parameters. The TLBO simply requires common
regulating parameters, and as a result, it has attracted the attention and acceptance of the research community.

The TLBO algorithm is a teaching-learning approach that is derived from the phenomena of the teacher's
influence on the learners' results. It represents two phases for the purpose of learning:

 From the teacher (called as teacher phase)

 From the interacting of learners with another learners (called learner phase)

The learning set is considered as a population in this approach, and non-identical subjects provided to learners are
treated as divergent design factors for the optimization, and the learner's output is regarded as the issue's 'fitness value'
[12].

The best solution obtained from the entire population is referred to as a teacher. The variables of design are those
that are incorporated in an objective function in order to arrive at the best solution, which is the best value obtained from
the objective function. This is a simple method that creates the simulation process's class room process. It frequently
necessitates common control parameters rather than algorithm-specific control criteria [12]. [13] is the enhanced version of
TLBO. This approach does not require gradient information or an error function. The major benefit of TLBO is that it does
not need any parameter tweaking. The artistic depiction of the procedure is represented in Figure 1.

Impact Factor (JCC): 7.0125 NAAS Rating 2.96


Microgrid BSS Scheduling Using Teaching Learning Based Optimization Algorithm 9

Figure 1: Schematic Diagram Depicting the Order of Proposed Methodology

CASE STUDY

Over a 24-hour period, we acquired the hourly load (Xt), produced energy (Gt), and grid pricing (At) profiles from [1].
Microgrids of size 8 houses with a total battery capacity of 43.44kWh are included in this data set. In this simulation, we
assume that in each microgrid, the cumulative batteries constitute a centralized BSS. We also assume that each BSS's SoC
is 30% at the beginning of the day. DD, the greatest depth of discharge, has a value of -23. (which means SoC can fall by at
most 23 percent in an hour interval). DC, the greatest depth of charge, is also +23. As a result, the search space has a total
of 100(DD - DC+ 1)24 = 470024≈2288 items.

Parameters Applied

Table 1: Parameter Values for PSO and TLBO


PSO Parameters TLBO Parameters
Parameter Value Parameter Value
50
Number of iterations
1
Number of design variables
Population Termination Criteria 50
24
Inertia of weight, w Number of Subjects 1
0.88-0.38
Constant C1 Number of Students 24
2
Constant C2 Rand 0-1
2
Rand 1 Upper bound 90
0-1
Rand 2 Lower bound 10
0-1
Upper bound
90
Lower bound
10

Figure 2: Data Forecast for 8 Homes in 24 Hours.

www.iaset.us [email protected]
10 Yendru Sravani, Dr. M. Veera Kumrai & N. Rama Narayana

Cost Reduction Using PSO

The PSO method is used to optimize a cost function with provided restrictions, as shown in Eq. 5, and the outcome is
shown in Figure 3. In Figure 3, the greatest cost is 4995.60 cents, while the lowest cost is 4688.68 cents.

When comparing the PSO and NPBA algorithms, the PSO method costs less, whereas the NPBA cos is 5471.22
cents [1]. It compares its efficacy to that of NPBA and calculates the improvement percentage. PSO hourly SoC is shown
in Figure 4. The SOC using PSO is 70 %, indicating that the battery is working well and is in good health.

Figure 3: Cost Reduction Using PSO.

Figure 4: Hourly SoC by PSO.

Cost Reduction Using TLBO

The TLBO method was used to optimize a cost function with provided restrictions, and the highest cost was 4775.63 cents
and the minimum cost was 4407.53 cents, as shown in fig. 5. When compared to NPBA and PSO, the cost of the TLBO
algorithm is significantly lower.

It compares its efficacy to NPBA and PSO and calculates the gain percentage. It compares its efficacy to NPBA
and PSO and calculates the gain percentage. The hourly SoC for TLBO is 79 percent while we applied the TLBO
algorithm, as shown in Figure 6. It clearly demonstrates the superiority of the TLBO algorithm over the PSO method.

Impact Factor (JCC): 7.0125 NAAS Rating 2.96


Microgrid BSS Scheduling Using Teaching Learning Based Optimization Algorithm 11

Figure 5: Cost Reduction Using TLBO.

Figure 6: Hourly SoC by TLBO.

COMPARISON ANALYSIS

The NPBA cost for eight houses is 5471.22 cents. With this in mind, The outcomes shown above are for an 8-home
microgrid. The algorithms were then run on the 16-home and 32-home microgrids, with similar results, as shown in Table
2. It's worth noting that the last column, Avg. Above all three data sets, Cost Reduction is an unweighted average of the
percentage improvement over NPBA delivered by the individual method. It's worth noting that the performance of our
suggested algorithms is highly reliant on the parameters used. Parameter settings that are effective for our data sets may or
may not be effective for other data sets. As a result, it's crucial to experiment with various parameter values until you find
one that works well for the data set in question.

Table 2: Comparison Analysis


Algorithm Cost Obtained Cost Reduction %
PSO 4688.68 14.31
8 Homes
TLBO 4407.53 19.45
PSO 9373.69 14.34
16 Homes
TLBO 8810.2 19.49
PSO 18754.92 13.9
32 Homes
TLBO 17724.22 19.02

CONCLUSIONS

The findings of the entire article are presented in this section. Real-time pricing, sometimes referred to as dynamic pricing,
is a utility rate structure in which the per-kWh fee fluctuates every hour based on the utility's current production prices.
Real-time pricing makes retail energy prices more expensive during peak hours than during shoulder and off-peak hours

www.iaset.us [email protected]
12 Yendru Sravani, Dr. M. Veera Kumrai & N. Rama Narayana

since peaking facilities are more expensive to run than base load plants. The cost is reduced using the PSO and TLBO
algorithms, with PSO achieving a cost of 4688.98 cents. The price is 4407.53 cents according to TLBO. The cost reduction
percentages are calculated using the basic NPBA cost as a benchmark, which is 5471.22 cents. Using PSO results in a
14.31 percent cost decrease. The cost savings from utilizing TLBO is 19.45 percent. In comparison to the PSO, it is
apparent that TLBO has a higher cost reduction percentage. We then applied the algorithms to forecast data obtained from
[19].As a result, it is obvious that the TLBO method is more productive than the PSO algorithm. The greatest SOC attained
with TLBO is 79%, while the PSO is 70%. This clearly demonstrates the precision of the TLBO algorithm when used in a
microgrid to determine the battery's level of charge. When compared to PSO, the efficacy of TLBO may be expressed as
the number of iterations considered being 100, which results in higher performance. PSO and TLBO's performance will be
almost identical after 100 cycles, making comparison impossible.

REFERENCES

1. Raghavan, P. Maan and A. K. B. Shenoy, "Optimization of Day-Ahead Energy Storage System Scheduling in
Microgrid Using Genetic Algorithm and Particle Swarm Optimization," in IEEE Access, vol. 8, pp. 173068-
173078, 2020, doi: 10.1109/ACCESS.2020.3025673.

2. M. I. Henderson, D. Novosel, and M. L. Crow, ``Electric power grid modernization trends, challenges, and
opportunities,'' in Proc. IEEE Technol. Trend Paper, 2017, pp. 1_17.

3. X. Fang, S. Misra, G. Xue, and D. Yang, ``Smart grid_The new and improved power grid: A survey,'' IEEE
Commun. Surveys Tuts., vol. 14, no. 4, pp. 944_980, 4th Quart., 2012.

4. H. Farhangi, ``The path of the smart grid,'' IEEE Power Energy Mag., vol. 8, no. 1, pp. 18_28, Jan. 2010.

5. Banerji, D. Sen, A. K. Bera, D. Ray, D. Paul, A. Bhakat, and S. K. Biswas, ``Microgrid: A review,'' in Proc. IEEE
Global HumanitarianTechnol. Conf., South Asia Satell. (GHTC-SAS), Aug. 2013, pp. 27_35.

6. D. T. Ton and M. A. Smith, ``The U.S. department of Energy's microgrid initiative,'' Electr. J., vol. 25, no. 8, pp.
84_94, Oct. 2012.

7. Zhou, K. Qian, M. Allan, and W. Zhou, ``Modeling of the cost of EV battery wear due to V2G application in
power systems,'' IEEE Trans.Energy Convers., vol. 26, no. 4, pp. 1041_1050, Dec. 2011.

8. P. Sanghvi, ``Flexible strategies for load/demand management using dynamic pricing,'' IEEE Trans. Power Syst.,
vol. 4, no. 1, pp. 83_93, Feb. 1989.

9. T. Logenthiran, D. Srinivasan, and E. Phyu, ``Particle swarm optimization for demand side management in smart
grid,'' in Proc. IEEE Innov. SmartGrid Technol. Asia (ISGT ASIA), Nov. 2015, pp. 1_6.

10. T. Logenthiran, D. Srinivasan, and T. Z. Shun, ``Demand side management in smart grid using heuristic
optimization,'' IEEE Trans. Smart Grid, vol. 3, no. 3, pp. 1244_1252, Sep. 2012.

11. Setlhaolo, X. Xia, and J. Zhang, ``Optimal scheduling of household appliances for demand response,'' Electr.
Power Syst. Res., vol. 116, pp. 24_28, Nov. 2014.

12. “Teaching-Learning-Based Optimization (TLBO) Algorithm And Its Engineering Applications”,Springer


International Publishing, Switzerland, 2016, DOI:10.1007/978-3-319-22732-0.

Impact Factor (JCC): 7.0125 NAAS Rating 2.96


Microgrid BSS Scheduling Using Teaching Learning Based Optimization Algorithm 13

13. Bingchuan, Wang & Li, Han-Xiong& Feng, Yun. (2018),“An Improved Teaching-Learning-Based Optimization
for Constrained Evolutionary Optimization”,Information Sciences. 456. 10.1016/j.ins.2018.04.083.

14. Y. Yang, J. Wu, Y. Chen, and C. Li, ``A new strategy for short-term load forecasting,'' Abstract Appl. Anal., vol.
2013, 2013, Art. no. 208964, doi: 10.1155/2013/208964.

15. S. Skolthanarat, U. Lewlomphaisarl, and K. Tungpimolrut, ``Short-term load forecasting algorithm and
optimization in smart grid operations and planning,'' in Proc. IEEE Conf. Technol. Sustainability (SusTech), Jul.
2014, pp. 165_171.

16. S. Dutta, Y. Li, A. Venkataraman, L. M. Costa, T. Jiang, R. Plana, P. Tordjman, F. H. Choo, C. F. Foo, and H. B.
Puttgen, ``Load and renewable energy forecasting for a microgrid using persistence technique,'' Energy
Procedia, vol. 143, pp. 617_622, Dec. 2017.

17. J. Zheng, C. Xu, Z. Zhang, and X. Li, ``Electric load forecasting in smart grids using Long-Short-Term-Memory
based recurrent neural network,'' in Proc. 51st Annu. Conf. Inf. Sci. Syst. (CISS), Mar. 2017, pp. 1_6.

18. Ahmad, N. Javaid, A. Mateen, M. Awais, and Z. A. Khan, ``Shortterm load forecasting in smart grids: An
intelligent modular approach,'' Energies, vol. 12, no. 1, p. 164, Jan. 2019.

19. Z. Garroussi, ‘‘Data for demand-side management of multiple homes,’’ Mendeley Data, V1, 2019, doi:
10.17632/hkyjg2spxf.1.

www.iaset.us [email protected]

You might also like