Smart Buildings : Use cases

Pierre Boislève
Master Internet of things
Institut Supérieur d’Électronique et du
Numérique
Nantes, France
pierre.boisleve@isen-ouest.yncrea.fr

Abstract—Smart buildings use sensor data to improve levels to be monitored and adjusted to ensure occupant
energy efficiency and improve performance. Different methods comfort while reducing energy consumption.
and techniques have been used to collect, analyze and use data. Light sensors: They measure the level of light inside the
Measurement sensors and building monitoring systems building. This data is used to regulate the lighting according
facilitate data collection, while statistical analysis and big data
techniques help identify energy consumption patterns Advanced
to the natural light available, which makes it possible to
control and optimization techniques are used based on sensor reduce the energy consumption linked to artificial lighting.
data to deliver real-time changes in temperature, ventilation,
and air conditioning parameters. Data-driven modeling and
B. Building management systems (BMS):
simulation techniques enable performance analysis and strategy
testing. The document highlights the importance of integrating BMS are computerized systems that collect and centralize
data-driven approaches in smart building design to achieve data from various building sensors. These systems make it
energy efficiency and sustainability. possible to monitor and control several aspects of energy
consumption, such as heating, air conditioning, lighting,
Keywords—Smart Buildings, Energy Management, Sensor ventilation, etc. BMSs facilitate the collection of real-time
Data, Data Analysis, Energy Efficiency data, the analysis of energy consumption patterns and the
automation of adjustments to optimize energy efficiency.
They have several specific features and techniques.
I. DATA GATHERING Data acquisition: The BMS collect data from measurement
Collecting, analyzing, and utilizing sensor data to improve sensors distributed throughout the building. This data
energy consumption in buildings has become increasingly includes measurements of energy consumption, temperature,
important in the pursuit of sustainable and efficient practices. humidity, light, air quality, etc. BMSs use standard
In the EU-28, fossil fuels are responsible for almost 75% of communication protocols such as BACnet, Modbus or Lon
the total energy consumption. Buildings use more than two- Works to retrieve sensor data and consolidate it into a
thirds of their energy consume for space heating (37%), water centralized database.
heating (12%), space cooling (10%) and lighting (9%) Control and automation: BMSs make it possible to control
purposes. To achieve this, various methods and techniques are and regulate heating, air conditioning, lighting, ventilation
employed to collect data from sensors, allowing for a systems, etc., based on the data collected. Using advanced
comprehensive understanding of energy usage patterns within
algorithms, BMS makes real-time decisions to optimize
a building. This information serves as the foundation for
equipment settings and minimize energy consumption. For
implementing effective energy management strategies and
optimizing resource utilization. In this context, the first point example, the BMS can automatically adjust the set
of focus revolves around the methods and techniques used for temperature, lighting power or airflow according to actual
data collection, enabling the acquisition of valuable insights needs, occupancy schedules, weather conditions, etc.
into energy consumption within buildings. Data analysis: BMSs use data analysis techniques such as
machine learning and statistical analysis to extract relevant
information from collected data. For example, predictive
A. Measurement sensors: models can be used to estimate future energy consumption
based on historical data and observed trends. To sum up, data
Measurement sensors play an essential role in collecting
analysis helps identify consumption patterns, bottlenecks,
data for improving the energy consumption of buildings.
anomalies, and optimization opportunities.
They make it possible to collect precise information on
Advanced User Interfaces: BMSs offer user-friendly user
various energy parameters, which provides a solid basis for
interfaces that allow building managers to easily view and
the analysis and optimization of the energy performance of
control energy performance. Dashboards and interactive
buildings.
graphs display real-time data, consumption histories, energy
Energy consumption sensors: These sensors measure the
efficiency reports, and more. Managers can monitor and
amount of energy consumed by the various equipment in the manage energy performance, set energy savings goals and
building, such as lighting, heating, air conditioning, assess the impact of efficiency measures put in place.
household appliances, etc. They provide accurate data on
energy consumption in real time.
Temperature sensors: They measure the temperature at II. DATA ANALYSIS
different places in the building, which makes it possible to
detect temperature variations and to optimize heating and air With the rise of digital technologies and the massive
conditioning systems accordingly. collection of data, it has become crucial to develop effective
Humidity sensors: These sensors measure the humidity analytical methods and techniques to extract valuable insights
level in the building. The data collected allows humidity

1
from this data. Data analysis aims to uncover hidden trends, developed and used in this context. Here we will look at some
patterns, and relationships within data sets, which helps to of the most used approaches.
make informed decisions, predict future events, and optimize
performance. 1) Simulated Annealing
The simulated annealing algorithm is widely used in the
optimization of predictive cyber-physical systems. Inspired
A. Pre-processing
by the annealing process used in metallurgy, this algorithm
Data pre-processing is a crucial step in the development explores the search space by allowing moves towards sub-
of predictive cyber-physical systems for smart buildings. optimal solutions with some probability. This avoids getting
This step consists of converting, cleaning and standardizing stuck in local minima and eventually reaching the optimal
the data collected to make them usable for analysis and global solution. The simulated annealing algorithm is suitable
decision-making. In this section, we will look at the different for nonlinear optimization problems and can be used to solve
approaches used to pre-process data in these systems. complex problems such as optimizing energy consumption in
buildings.
1) Theoretical approaches
Theoretical approaches to data pre-processing rely on 2) Particle Swarm Optimization (PSO)
mathematical and statistical models to analyze and filter data. The Particle Swarm Optimization (PSO) is another
These models are usually developed by experts in the field popular heuristic. It relies on a population (denoted swarm)
and require regular maintenance to ensure their accuracy. of candidate solutions (particles). The heuristic is based on a
They are often based on specific domain knowledge and gravitational metaphor to iteratively update the particles
assumptions about the relationships between different according to simple rules of attraction and inertia [1]. Each
variables. These approaches are used when the theoretical particle moves through space according to its own best-
models are well established and the data are relatively stable. known solution and the global best solution found by the
For example, in the field of building energy management, whole group. The goal is to find the best solution by
theoretical models based on the principles of converging the particles to a region of space that contains the
thermodynamics and building physics are used to estimate optimal solution. PSO is often used in cyber-physical systems
energy consumption. These models consider factors such as to optimize parameters such as heating, ventilation, and air
outdoor temperature, building size, heating and cooling conditioning (HVAC) scheduling in buildings.
system characteristics, etc. The data collected is compared to
the values predicted by the theoretical model, and 3) Other optimization techniques
adjustments are made in the event of a significant deviation. Besides simulated annealing and particle swarm
optimization, there are other optimization techniques used in
2) Data-driven approaches predictive cyber-physical systems of smart buildings. The
Data-driven approaches use machine learning and data genetic algorithm is an approach inspired by the theory of
mining techniques to pre-process data. These approaches are biological evolution. It uses operations of reproduction,
often used when the theoretical models are not precise crossover and mutation to generate a population of candidate
enough or when the data is complex and variable. They help solutions and improve them over generations. This allows to
uncover hidden relationships between variables, identify efficiently explore the search space and find optimal solutions
patterns or anomalies, and generate more accurate prediction for complex problems.
models.
In predictive cyber-physical systems in smart buildings, Other techniques, such as the firefly algorithm and the
the data collected can be extremely diverse, ranging from cuckoo search, have also been applied in this field. The firefly
temperature and humidity measurements to occupant algorithm is based on the attraction and repulsion behavior of
information such as comfort preferences and behavioral fireflies, while the cuckoo search is inspired by the egg-laying
habits. To sum up, these approaches train regression models, strategy of certain parasitic birds. These approaches offer
such as Neural Networks, using observed building specific advantages depending on the context and the type of
management system (BMS) data. optimization problem.
However, one of the major challenges of data-driven
approaches is the management of environmental changes.
Occupant behaviors, occupancy patterns, and even seasons III. DATA UTILIZATION
can change abruptly, rendering pre-processing models
obsolete. Therefore, it is necessary to regularly update the
computational representation models to adapt to these A. The three-layered architecture
changes. The three-layered architecture is commonly found in
building automation. It plays a crucial role in the control and
management of buildings. We can find three different layers
B. Optimization
in which we are going to delve deeper.
Optimization plays a crucial role in predictive cyber-
physical systems of smart buildings, as it aims to improve 1) Field Level
control decisions by searching for the most efficient This layer encompasses the physical components of the
solutions. Different optimization techniques have been building automation system, including sensors and actuation
devices. Sensors collect data related to various parameters

2
such as temperature, humidity, occupancy, and energy IV. USE CASES
consumption. Actuators, on the other hand, receive control
commands and perform physical actions, such as adjusting
lighting levels, opening or closing valves, or activating A. Multi-objective optimization for thermal mass model
HVAC systems. The field level is responsible for capturing predictive control in small and medium size commercial
real-time data and interacting directly with the building's buildings under summer weather conditions [2]
infrastructure. [1] This study focuses on optimizing the operation of HVAC
systems in six small to medium-sized commercial buildings.
2) Automation Level The aim of the study is to achieve energy cost savings by
The automation level acts as an intermediary between the improving the control strategy of the HVAC systems.
field level and the management level. It consists of controllers In this study, the researchers used a precooling strategy
that receive input from sensors and make control decisions based on five-day weather data in August. The precooling
based on predefined algorithms and set points. These strategy involved lowering the temperature of the building
controllers implement control loops to regulate and optimize during off-peak hours when electricity prices are typically
building systems, ensuring that they operate within specified lower, and then maintaining a slightly higher temperature
parameters and meet desired performance criteria. The during peak hours to reduce energy consumption.
automation level is responsible for executing control The researchers compared the energy cost savings
strategies, adjusting set-points, and coordinating the actions achieved through the precooling strategy with the energy
of different systems within the building. consumption and costs associated with a conventional control
strategy for HVAC systems. The conventional control
3) Management Level strategy typically maintains a constant temperature
The management level represents the highest layer of the throughout the day, regardless of the electricity prices or
architecture and is responsible for overseeing and outside weather conditions.
coordinating the entire building automation system. It The results of the study demonstrated significant energy
typically consists of a central computer or server hosting the cost savings ranging from 20% to 60% when implementing
Building Management System (BMS). The BMS serves as the precooling strategy [2]. The savings were observed across
the control hub, providing a user interface for operators to the six commercial buildings, which varied in size and
monitor and manage the building's systems. It allows occupancy patterns.
operators to configure set-points, create schedules, set rules, By using five-day weather data in August [2], the
and access historical data for analysis and decision-making. researchers were able to optimize the precooling strategy and
The management level also facilitates integration with align it with the expected weather conditions. This approach
external systems, such as energy management systems or allowed the HVAC systems to take advantage of the lower
weather forecast services, to optimize energy consumption electricity prices during off-peak hours, thereby reducing
and overall building performance. energy costs while still maintaining a comfortable indoor
environment during peak hours.
The study emphasizes the importance of considering
external factors, such as weather conditions and electricity
prices, when optimizing the operation of HVAC systems in
commercial buildings. By integrating these factors into the
control strategy, significant energy cost savings can be
achieved without compromising occupant comfort.

B. Real-time Energy Control Approach for Smart Home

Energy Management System [3]
This study focuses on demand response (DR)
management by integrating a building's peak demand with
DR incentives into a single equation [1]. The study aims to
reduce electricity costs by optimizing the participation of
appliances in demand response programs.
Demand response programs involve modifying electricity
consumption patterns in response to grid conditions or pricing
signals. By shifting or reducing electricity usage during peak
demand periods, consumers can help alleviate strain on the
grid and potentially earn incentives or lower electricity rates.
In this study, rolling optimization of a multi-hour
Fig. 1. Integration of the predictive CPS concepts with the common three prediction horizon with minute-based real-time control
layer BMS structure. [1]
strategies and fuzzy logic techniques are used to manage
demand response in buildings. The researchers propose a
formulation that combines the building's peak demand with
the incentives offered by the demand response program.

3
 The approach involves continuously monitoring the
building's energy usage and the current incentives offered by
the demand response program. Based on this information, the
optimization algorithm determines the optimal scheduling of
appliances to minimize electricity costs while considering the
building's comfort and operational constraints.
The study demonstrates significant cost reduction. The
results show that the proposed approach can achieve
electricity cost savings ranging from 16% to 19% compared
to a default strategy [3].

REFERENCES
[1] M. Schmidt and C. Åhlund, “Smart buildings as Cyber-Physical
Systems: Data-driven predictive control strategies for energy
efficiency,” Renewable and Sustainable Energy Reviews, vol. 90, pp.
742–756, Jul. 2018, doi: https://doi.org/10.1016/j.rser.2018.04.013.
[2] X. Li and A. Malkawi, “Multi-objective optimization for thermal mass
model predictive control in small and medium size commercial
buildings under summer weather conditions,” Energy, vol. 112, pp.
1194–1206, Oct. 2016, doi:
https://doi.org/10.1016/j.energy.2016.07.021.
[3] S. Zhou, Z. Wu, J. Li, and X. Zhang, “Real-time Energy Control
Approach for Smart Home Energy Management System,” Electric
Power Components and Systems, vol. 42, no. 3–4, pp. 315–326, Feb.
2014, doi: https://doi.org/10.1080/15325008.2013.862322.