Babylonian Journal of Machine Learning
Vol.2024, pp. 80 – 88
DOI: https://doi.org/10.58496/BJML/2024/008; ISSN: 3006–5429
https://mesopotamian.press/journals/index.php/BJML
Research Article
Enhancing Energy Efficiency With Smart Building Energy Management System
Using Machine Learning and IOT
M.Sahaya Sheela1* , S.Gopalakrishnan2 , I.Parvin Begum3 , J. Jasmine Hephzipah4 , M Gopianand5 , D. Harika6
1
Department of Department of Electronics and Communication Engineering , Vel Tech Rangarajan Dr. Sagunthala R&D Institute of Science and
Technology, Chennai, Tamil Nadu, India.
2
Department of Information Technology, Hindustan Institute of Technology and Science, Kelambakkam, Tamil Nadu 603103, India.
3
Department of Computer applications , B.S.Abdur Rahman Crescent Institute of Science and Technology , Va ndalur, Tamil Nadu 600048, India.
4
Department of Electronics and Communication Engineering, R.M.K. Engineering College, Kavaraipettai, Tamil Nadu 601206, India,
5
Department of Computer Applications, PSNA College of Engineering and Technology, Dindigul 624622.Tamil Nadu, India.
6
Department of Electronics and Communication Engineering, Mohan Babu University (Erstwhile Sree Vidyanikethan Engineering College),
Tirupati-517102.India.
A R TICLE INFO A BS TR ACT
Article History
Received 21 Mar 2024 The energy management system designed on the networking platform has been interfaced with
Accepted 19 May 2024
controller to control the electrical device using the Wireless communication has been used as the most
Published 11 Jun 2024
reliable and efficient technology in short-range communication. In this method IoT-based energy
Keywords management could significantly contribute to energy conservation of home appliances device. This
Power Supply model analyses an IoT-based smart energy meter that automatically tracks residential energy
Voltage Sensor consumption using current and voltage sensors. Input values senses unit that detects and controls the
electrical devices used for daily actions. The ESP32 is used due to its built-in Wi-Fi facility, allowing
Esp32 controller
data collection and exchange from electronic hardware to a cloud platform. The virtual android app
Blynk application
displays the value of voltage, current, power, and unit consumed on a mobile screen, enhancing the
Current Sensor. efficiency of the system. The developed coding system to enhance system performance and provide
more accurate results and ESP32 controller to interface non-invasive CT and voltage sensors,
delivering data to a Blynk server over the internet. Model show the system accurately records voltage,
current, dynamic power, and increasing power consumption and outcome accordingly, the home
concerned person can turn ON/OFF the device based on such information if customer based user
information.
1. INTRODUCTION
Energy management refers to the automation of the house or its activities involve the control of lights, fans, appliances,
and other systems to improve convenience, comfort, electricity savings, and security. The concept of energy management
has grown over many existences, and devices have recently entered the market; however, no personal solution has yet
broken through to become widespread.
81 Sheela et al, Babylonian Journal of Machine Learning Vol.2024, 80–88
Fig. 1. Application of Smart Home management
Energy management also benefits the elderly and disabled who have moved into their homes by eliminating the need to
go from one spot to another only to turn on or off appliances, open the door, and so forth. This can also provide a remote
interface to home kitchen appliances or the automation system over the internet, allowing control
A smart building is one that uses technology to exchange information about what occurs in it at its basic level. Construct
connections between systems to maximize the efficiency of the structure. Then, these data are employed to automate a
range of procedures, including heating and ventilation, climate control, and safety. Smart workplaces and smart homes,
libraries, healthcare facilities, and other smart hospitals and a variety of other SBs that can offer automated services that
can bring a variety of value assistance and services (such as reducing energy waste) to safeguard the tenants’
convenience, health, and safety.
1.1 Objective
To integrate all devices with networking communication and device-to-device communication into a discrete-time
model
To create a system to plan all building electrical equipment in order to achieve energy savings in accordance with
real-time voltage and current measurements for consumers.
To track and estimate bills and increase consumer understanding of energy consumption among residential and
commercial users.
To achieve this, a monitor device shows the real-time estimated energy used by each connected component and
the real-time estimated bill of the entire consumption.
2. LITERATURE SURVEY
Machine learning can be used to automate numerous operations, Smart buildings make it easier to manage tasks such as
temperature, safety, and maintenance using computers and mobile devices. The Internet of Things (IoT) is used by smart
buildings to link various building systems. Smart buildings are becoming vital parts of larger system integration as the
Internet of Things develops. The Internet of Things (IoT) is essential for smart buildings and provides features that boost
82 Sheela et al, Babylonian Journal of Machine Learning Vol.2024, 80–88
security for people by using potent tech-based critical approaches. This study presents the HEMS-IoT smart home energy
management system, which is built on big data and machine learning for home comfort, safety, and energy savings. [1].
Because of advancements in energy conversion, new communication techniques, and information technology, smart
houses are becoming a reality. The Internet of Things (IoT) is an amalgam of diverse technologies from several
application domains that are employed to connect objects over the Internet, enabling the identification, observation, and
management of multiple objects from a distance. Options that promise to improve energy efficiency include big data,
machine learning, and energy management systems integrated with the Internet of Things. Blynk API and the J48
machine learning algorithm to comprehend user behaviors and energy usage. [2].
Demand-oriented systems frequently include smart Internet of Things (IoT)-capable devices in an ecosystem designed to
maximize energy efficiency and have been proposed for smart home energy management systems. However, these
technologies are only applicable to already-intelligent devices and are not suitable for many locations where most
appliances are not yet IoT-capable. The argument for paying consideration to non-smart equipment and providing a way
for incorporating them into the energy-saving environment of the IoT [3].
The energy consumption is statistics on maximum demand's hourly consumption collected at the smart commercial
tenants that have been chosen for Internet of Things (IoT)-enabled construction has been analyzed using a arithmetic
technique using the formula, the mean, and the variance. K-nearest Neighbor On the consumption data, the (k-NN)
approach has been used. Using three distinct values of the closest distance for the prediction method. The predicted data
were divided into various testing and training ratios of 70% and 30%, respectively [4].
Therefore, a continuous energy system is required to create a national performance indicator. Thus, a better system
known as the smart grid is installed to improve the energy management system. Addressing demand-side management in
smart buildings promotes effective energy management. The use of optimization techniques, such as genetic algorithms,
helps to reduce the added process of energy consumption in smart buildings. This is accomplished using an optimization
technique and machine learning. This facilitates efficient appliance scheduling. The scheduling of priorities is used to
accomplish this. The total performance was measured using sensors and the Internet of Things. The act of detecting
exterior physical parameters such as temperature, humidity, wetness, and heat [5].
Using gradient boosting machine techniques, such as the Light Gradient Boosting Machine (Light GBM), Category
Boosting (Cat Boost), and Adaptive Boosting (Ada Boost), the energy consumption of certain public buildings was
estimated. The key predictors developed by each technique were compared, and the Light Gradient Boosting Machine
approach (Light GBM) produced the best-performing model with the lowest root mean squared error of 1.119. The
experiment employed R-squared, mean absolute error, and root mean squared error as assessment measures. As part of
the idea of "smart cities," the model may be used to create an intelligent machine learning-based energy management
system for public buildings. [6].
Due to the rising population and their aspirations for comfort, there has been a tremendous rise in global energy
consumption over the past few decades. Building energy management has become a crucial topic in today’s society for
decreasing energy waste because buildings account for 40% of the world’s energy. The deployment of energy-efficient
management systems requires the ability to predict the amount of energy consumed in SB and plan for that use. Approach
for calculating the energy consumption of buildings and providing recommendations for reducing energy consumption
and costs. Information such as temperature, humidity, and air quality is used by the systems. Machine learning is used to
analyze these data and predict the number of energy smart buildings will use [7].
Due to the increased use of BLE technology, conventional ZigBee-based energy efficiency management solutions are
less flexible in smart homes. This study provides a detailed comparison of the energy-saving management capabilities of
ZigBee and BLE in smart homes. A BLE-based smart home energy management system is recommended. The BLE
technology’s fast transmission rate and high degree of interaction are taken advantage of by this system. Additionally, the
study creates new concepts like "rules-based" smart home energy management systems, "smart app terminals for energy
efficiency management," and "smart home." [8].
Energy specialists and managers are thus looking at these building energy demand profiles to optimize and boost the
energy efficiency in the structure itself first and subsequently in the grid. This paper aims to present various algorithms
that predict the electrical power use of structures, a case study of a hotel building in Shanghai, and to describe the
approach needed and the comparative study between them using machine learning algorithms in Rapid Miner software,
such as support vector machine (SVM), artificial neural network (ANN), selection tree (DT), and random forest (RF). [9].
A method for a smart residential energy management system that employs machine learning is suggested in this study.
With rooftop solar and locally energized local storage, the proposed system shifts pre-prioritized prospective loads
between the grid and residential premises using machine learning algorithms. Using a load-shifting algorithm, it is also
possible to reduce electricity expenditure in residential buildings while maintaining a steady power supply. The system
83 Sheela et al, Babylonian Journal of Machine Learning Vol.2024, 80–88
also includes features that predict average solar power using artificial neural networks and optimize solar power
generation and energy storage using reinforcement learning. In the end, residential premises have less reliance on the grid
[10].
Using cutting-edge technologies, this study focuses on building energy management and conservation. Machine learning
and the Internet of Things (IoT) are used to automate building equipment and track energy consumption. The work
discusses energy management and conservation using advanced IOT technologies such as sensors and actuators in
buildings. For efficient use of electricity in the building, the appliances are sensed using various sensors and monitored
by Arduino. The Smart Home Bluetooth-controlled Android app was created using MIT App Inventor. Local automation
is performed using Bluetooth and the MIT App Inventor to monitor and manage the building’s electrical equipment from
handheld devices. Customers are observed using electricity carelessly [11].
Building management systems need to be able to anticipate occupancy in real time with greater accuracy as Internet of
Things (IoT) sensors become more sophisticated and are used in smart building automation applications. This saves
power and maintains the comfort of the people. The experimental conclusions show that GRU outclasses the LSTM
network because it uses 13.57% less power and has a reduced error of 1.21%.[12].
This study develops a novel energy-efficient thermal comfort model for environmental conditioning systems in smart
structures, with the goal of reducing energy consumption and improving occupants' thermal comfort. Our provide a
system architecture for a thermal comfort system that would reduce energy consumption and increase occupant comfort
while taking into consideration outliers in groups like people of different ages. Using the ASHRAE RP-884 database,
data were gathered during the sensing phase. Then, using various machine learning methods, we apply a machine
learning approach to forecast the Thermal Sensation Vote (TSV) of the thermal comfort model [13] - [14].
Energy consumption forecasts are useful for consumer expenditure planning, comfort control in smart buildings, and
energy budgeting for smart grids. Energy management is a fundamental requirement for building management systems in
smart buildings. To achieve this, energy forecasting is essential. In general, machine learning techniques are used to
predict energy use. However, massive amounts of data are required for machine learning algorithms to perform
successfully. The collection of these data from data slaveholders may influence privacy breaches. A distributed system
framework called federated learning can reduce privacy concerns. Therefore, federated learning has been developed for
mobile edge devices such as automobiles, smartphones, and other mobile devices [15].
An efficient smart grid can integrate any type of generating source into the system, offering end users dependable, useful,
intact, and high-quality electricity. The most important use case for the smart grid is the Home Energy Management
System (HEMS). Energy forecasting is crucial in HEMS because of the non-uniform and extremely dynamic loading
circumstances and the changeable generation availability. This article provides a quick introduction to load forecasting
and HEMS approaches. A unique approach based on Long Short-Term Memory (LSTM) and Deep Neural Network
(DNN) for estimating periodic power usage. The load is predicted for the next seven days. [16].
3. MATERIALS AND METHODS
The method uses the wireless input signal that is received from the Android handset to manage electrical loads. It gets
harder for the elderly and disabled to continually turn on and off electricity switches by manually. A smart house is a
structure with advanced technology installed that enables communication between several systems and appliances.
A computer-aided system that operates automatically and is set up by a computer in a building or house is known as
smart home lighting management. One way to give control orders is to touch a button on the equipment, or users may use
automated systems that can sense their surroundings on their own. The user may keep an eye on the condition of the
house in real time data to the usage of remote monitoring devices. In the design of electronics and power supplies,
converting AC (alternating current) to DC (direct current) at a particular voltage is a frequent problem. You can use a
power supply or converter to change AC 230 volts to DC 5 volts. DC 5v to connect the esp32 controller.
84 Sheela et al, Babylonian Journal of Machine Learning Vol.2024, 80–88
POWER
SUPPLY
(IOT)
VOLTAGE BLYNK
SENSOR ESP32
MICROCONTROLL
ER
LCD DISPLAY
CURRENT
SENOSR
Relay Relay
Control Control
Fan Control Light Control
Fig. 2. Block diagram of Proposed Energy Management
A pre-made AC to DC power supply with the required specs is available for purchase. These power supplies are widely
accessible and are offered in a variety of configurations, such as wall adapters, desktop power supplies, and DIN rail-
mounted devices. Make sure that the power supply you select has a 5V DC output rating and an input rating that is
compatible with your AC voltage (in this case, 230V AC). If you only need a 5V DC for charging or powering USB
devices, you can use a USB power adapter. These adapters typically accept a wide range of input voltages, including
230V AC, and provide a 5V DC output via a USB port.
3.1 Esp32 Controller
The ESP32 is a versatile controller that can connect to Wi-Fi or other communication protocols, making it suitable for
IoT applications. Numerous tools and technologies to track, manage, and optimize a building's energy use for
sustainability and efficiency. This system can make use of an ESP32 controller to manage energy-related operations,
interface with sensors and actuators, and gather data. The main elements and procedures for building a BEMS using an
ESP32 controller are listed. Various sensors can be used to monitor parameters like temperature, humidity, occupancy,
light levels, and energy consumption. Examples include temperature sensors, occupancy sensors (PIR or ultrasonic), light
sensors, and current/voltage sensors for monitoring electrical loads.
3.2 Voltage Sensor
Fig. 3. Module of voltage sensor with example display output
85 Sheela et al, Babylonian Journal of Machine Learning Vol.2024, 80–88
A voltage sensor is a crucial device used in various applications, including industrial controls and power systems. This
type of sensor is often used with controllers, as they can easily measure changes in electromagnetic fields around them
using built-in analog-to-digital converters (ADCs). Voltage sensors are divided into two types: voltage-sensitive resistors
(VSRs) and operational amplifiers (op-amps). VSRs are used in electronic circuits to detect small voltage changes, like
battery life or solar panel output. Op-amps are used for tasks like motor driving or LED control, and are commonly found
in controllers like ESP32 Controller boards.
3.3 Current Sensor
Fig. 4. Current sensor Configuration
A current sensor measures the electric current passing through a conductor, converting it into a quantifiable output such
as voltage, current, or digital signal. These sensors are used for monitoring, control, or protection, and can measure large
currents due to their coreless structure. Magnetic current sensors detect the generated magnetic field around the current
path, allowing for the measurement of current. They are different from shunt approaches in that they may be isolated
inside the sensor itself and do not require an isolation amplifier or isolation ADC. This approach is appealing because it
detects current flowing to resistors with relatively low resistance standards, hence addressing the drawbacks of shunt
resistor isolation amplifier/isolation ADC.
3.4 Blynk IOT
Fig. 5. Blynk IOT application
Blynk is an Internet of Things platform that works with iOS and Android devices to enable internet-based remote control
of Node MCU. It enables users to work with widgets to construct graphical interfaces, or HMIs. Blynk can store,
envision, and display sensor data. It was created with the Internet of Things in mind. The platform is made up of three
main parts: the Blynk App, which lets users modify interfaces using pre-made widgets; the Blynk Server, which manages
all smartphone-to-hardware communication; and the Blynk Libraries, which facilitate command processing and
communication for common hardware platforms. Data is moved to the Blynk Cloud when a button is touched, and it then
somehow makes its way to the installed devices. The system is able to manage thousands of connections.
86 Sheela et al, Babylonian Journal of Machine Learning Vol.2024, 80–88
3.4 LCD Display
Fig. 6. Model calculation of LCD output
Liquid crystal display, (LCD) demonstration module found in many gadgets, including calculators and cell phones. Its
main applications are seven segments and multi-segment light-emitting diodes. LCDs are programmable and affordable
and provide countless options for unique animations and characters. The data register and command register are the two
registers found in a 16 2-LCD. The data register is set to '1', while the command register is set to '0'. Register select (RS)
is used to switch between registers. This system can multitask; therefore, it will take less time to control loads, regulate
loads, and keep an eye out for motion and fire. For the preservation of the user’s energy: People occasionally become
exhausted from their work and lose part of their vitality. Because this mechanism operates autonomously, humans do not
need to exert more effort.
4. RESULT AND DISCUSSION
The outcome of Internet of Things (IoT) for controlling smart houses by using sensors and the Blynk application to
control appliances in the house. Two loads a fan, a light are used as examples of this proposed work, through the internal
controls, a user may effortlessly operate these appliances due to our logical interface.
Over the internet, this embedded device will receive orders from users. Loads are managed using relays. The controller
operates these loads by processing commands from the user over the internet. As an outcome, the technology enables
effective online energy management. With the aid of the technology, the loads will be automatically controlled.
Execution: There are several parts to the Internet of Things (IoT) smart energy management system, including those that
follow: Sensor Unit: The voltage and current passing through the electric power lines are measured by the sensors in this
module. The embedded device receives the data about these sensors and processes it. Controller: This component
transfers information to the wireless part after processing it from the sensor unit.
U5
POWER SUPPLY CIRCUIT
3 2
VI VO
www.TheEngineeringProjects.com
TR1 BR1
ADJ
1
VDD
VEE
VSS
RW
C1
RS
D0
D1
D2
D3
D4
D5
D6
D7
E
BRIDGE 1uF
TRAN-2P2S
RV1
VOLTAGE SENSOR
100K
DC LOAD VIN
GND
RD2
RST
RL3(C1)
RELAY RL3
12V
ESP8266 Controller
BLYNK APP (IOT)
R3 Q3
BC547
10k
RD8
CURRENT SENSOR
Tx Rx
Fig. 7. Circuit diagram of smart energy management
87 Sheela et al, Babylonian Journal of Machine Learning Vol.2024, 80–88
Fig. 8. (A) Output in the LCD display (B) Output in the Blynk application
Figure 8. Shows the real-time electrical parameters like current, voltage, power, and the unit consumed in the LCD
display, and the parameters are shown on the mobile device through the Blynk IoT application for the portable
monitoring of the electrical parameters.
5. CONCLUSION
The smart energy monitoring system is automated, using an ESP32 controller to interface with a current transformer and
voltage sensor. The data is then retrieved via the blynk cloud for web clients. Experiments show the system effectively
monitors voltage and current, power and kilo watt hours. This model aims to design and implement a smart energy meter
system using an ESP32 controller in homes and public spaces, measuring voltage, current, energy, and power using a
voltage and current sensor. This system that measures both mains voltage and current used by a consumer.
The voltage sensor's calibration parameters allow for high accuracy. It can measure current up to 30 amps and voltages
up to 250 volts. The technological aspects of the ESP32 development are described on the same page, including Wi-Fi
802.The 11 b/g/n connection and 18 ADC channel with 12-bit resolution make it suitable for this IoT project. The
equipment measures voltage, current, power use, and hourly consumption. The measured values appear on either the
Blynk mobile application via Wi-Fi or the LCD display. The tool also offers a graph that shows electricity values over a
12-month period. This graph can be saved in CSV format for further examination or consumption.
5.1 Future scope
In future Internet of Things (IoT) refers to machine-to-machine communication. Additionally, an IoT-based energy
management system in which a controller receives readings from environmental indicators including temperature and
light intensity sensors. The Arduino controller is configured to regulate the appliance consumption in accordance with
detected readings. Along with regulating appliance use, each appliance's current draw is calculated using Hall Sensors,
and the additionally added remotely sent to a Raspberry Pi 3 via a Wi-Fi component.
Conflicts Of Interest
The paper's disclosure section confirms the author's lack of any conflicts of interest.
Funding
No financial grants or awards related to the research are disclosed in the paper, signifying a lack of external funding.
Acknowledgements
I wish to acknowledge the facilities provided by Publishing this Research article by “Centre for Networking and Cyber
Défense” (CNCD) - Centre for Excellence, Department of Information Technology, Hindustan Institute of Technology
and Science, Kelambakkam, Tamil Nadu -603103, India
88 Sheela et al, Babylonian Journal of Machine Learning Vol.2024, 80–88
References
[1] S. F. A. Shah and M. Iqbal, "The Role of Machine Learning and the Internet of Things in Smart Buildings for
Energy Efficiency," Faculty of Computer Science and Information Technology, Virtual University of Pakistan,
2022.
[2] I. Machorro-Cano and G. Alor-Hernández, "HEMS-IoT: A Big Data and Machine Learning-Based Smart Home
System for Energy Saving," Tecnológico Nacional de México, I. T. Orizaba, 2020.
[3] E. A. Affum, K. Agyeman-Prempeh, and A. Agyekum, "Smart Home Energy Management System based on the
Internet of Things (IoT)," Int. J. Adv. Comput. Sci. Appl., 2021.
[4] N. L. Mazlan and N. A. Ramli, "A Smart Building Energy Management using Internet of Things (IoT) and Machine
Learning," Electrical Engineering Section, Universiti Kuala Lumpur British Malaysian Institute, 2021.
[5] J. S., C. N. Ravi, M. Sujatha, S. S. Southry, J. Sundararajan, and C. V. K. Reddy, "Machine Learning and IoT based
Performance Improvement of Energy Efficiency in Smart Buildings," in Proc. ICSCDS, pp. 375-380, 2023.
[6] A. T., J. Li, O. Bamisile, and C. Ohakwe, "Machine Learning-Based System for Managing Energy Efficiency of
Public Buildings: An Approach towards Smart Cities," in Proc. AES, pp. 297-300, 2022.
[7] K. Kumar, B. A. G. Gopireddy, and P. Ranjana, "Energy Consumption in Smart Buildings using Machine
Learning," in Proc. ICICCS, pp. 413-418, 2023.
[8] L. J., "Design of Smart Home Energy Efficiency Management System Based on BLE Technology," in Proc.
MLBDBI, pp. 201-206, 2022.
[9] A. Maghraoui, E. F. -E. Hammouch, Y. Ledmaoui, and A. Chebak, "Smart Energy Management System: A
Comparative Study of Energy Consumption Prediction Algorithms for a Hotel Building," in Proc. GPECOM, pp.
529-534, 2022.
[10] J. R. Wijesingha, B. V. D. R. Hasanthi, I. P. D. Wijegunasinghe, M. K. Perera, and K. T. M. U. Hemapala, "Smart
Residential Energy Management System (REMS) Using Machine Learning," in Proc. ICCIKE, pp. 90-95, 2021.
[11] L. P. Raju, V. K. S. A. A. S. V. V., and B. V., "Building Energy Management and Conservation using Internet of
Things," in Proc. ICSCDS, pp. 970-974, 2022.
[12] N. Fatehi, A. Politis, L. Lin, M. Stobby, and M. H. Nazari, "Machine Learning based Occupant Behavior Prediction
in Smart Building to Improve Energy Efficiency," in Proc. ISGT, pp. 1-5, 2023.
[13] M. Abdulgader and F. Lashhab, "Energy-Efficient Thermal Comfort Control in Smart Buildings," in Proc. CCWC,
pp. 0022-0026, 2021.
[14] Z. Wu and W. Chu, "Sampling Strategy Analysis of Machine Learning Models for Energy Consumption
Prediction," in Proc. SEGE, pp. 77-81, 2021.
[15] S. V. Damari, K. Mittal, S. GVK, J. Bapat, and D. Das, "Privacy Enhanced Energy Prediction in Smart Building
using Federated Learning," in Proc. IEMTRONICS, pp. 1-6, 2021.
[16] M. Nutakki, M. S. M., and S. Mandava, "Energy Consumption Forecasting in Home Energy Management System
using Deep Learning Techniques," in Proc. INDICON, pp. 1-6, 2022.
