Full Text
<article class="scholarly-article">
<h2>Introduction</h2>
<p>The global energy landscape is characterized by increasing demand and a pressing need for sustainable consumption practices. Buildings are significant contributors to this demand, accounting for approximately 40% of total energy consumption and 36% of CO2 emissions in the European Union, for example (Aguilar et al., 2021). Within these structures, Heating, Ventilation, and Air Conditioning (HVAC) systems are often the largest energy consumers, typically representing 40-60% of a building's total energy use (Nguyen & Aiello, 2012). Traditional HVAC control strategies, which rely predominantly on fixed schedules or simple thermostat setpoints, often fail to account for the dynamic and unpredictable nature of building occupancy. This disconnect leads to substantial energy waste, as systems may operate at full capacity in unoccupied or sparsely occupied spaces, simultaneously compromising occupant comfort in other areas (Hong et al., 2016; O’Brien et al., 2020).</p><p>The advent of smart building technologies, characterized by interconnected sensors, actuators, and intelligent control systems, offers unprecedented opportunities to optimize building performance. A critical component of this optimization lies in developing more responsive and efficient HVAC systems that can adapt in real-time to actual building usage. Real-time occupancy detection provides the necessary input, transforming static control into a dynamic, user-centric approach. By accurately knowing when and where occupants are present, HVAC systems can be precisely controlled to deliver heating or cooling only when and where it is needed, thereby minimizing energy consumption without sacrificing comfort (Labeodan et al., 2015; Esrafilian-Najafabadi & Haghighat, 2021).</p><p>However, merely detecting occupancy is insufficient for optimal control. The inherent thermal inertia of buildings means that immediate reactions to occupancy changes can be inefficient. This necessitates the integration of predictive control strategies, such as Model Predictive Control (MPC), which can anticipate future conditions—including occupancy patterns, external weather changes, and electricity pricing—to make proactive control decisions (Drgoňa et al., 2020; Gholamzadehmir et al., 2020). MPC offers a robust framework for optimizing HVAC operations over a future time horizon, considering various constraints and objectives, such as energy minimization and thermal comfort maximization (Dong & Lam, 2013; Avci et al., 2013).</p><p>This paper aims to provide a comprehensive overview of real-time occupancy detection and predictive control strategies for HVAC systems in smart office buildings. We explore the state-of-the-art in occupancy sensing, delve into the theoretical underpinnings and practical applications of MPC in the context of building energy management, and critically assess the benefits, challenges, and future directions of integrating these two powerful paradigms. The overarching goal is to present a rigorous examination of how these advanced control methodologies can contribute to more energy-efficient and occupant-friendly smart building environments, aligning with the broader objectives of sustainable development in the built environment.</p>
<h2>Literature Review</h2>
<p>The drive for energy efficiency in buildings has led to extensive research into advanced control strategies for HVAC systems, with a particular focus on leveraging real-time data. This section reviews the evolution of occupancy detection methods and predictive control techniques, culminating in their synergistic integration.</p><h4>Occupancy Detection Technologies</h4><p>Accurate and real-time occupancy information is fundamental to demand-driven HVAC control. Early approaches relied on scheduled occupancy profiles, which often proved inaccurate due to the stochastic nature of human presence (Hong et al., 2016). The need for dynamic data spurred the development of various sensor-based detection methods (Labeodan et al., 2015).</p><p>Traditional direct detection methods include Passive Infrared (PIR) sensors, which detect motion, and CO2 sensors, which infer occupancy based on human respiration. PIR sensors are cost-effective but can suffer from false negatives when occupants are stationary and false positives from non-human movement. CO2 sensors provide a good proxy for human presence but have a time lag and can be influenced by ventilation rates (Labeodan et al., 2015). More advanced direct methods involve camera-based systems, offering high accuracy but raising significant privacy concerns (Labeodan et al., 2016).</p><p>Indirect occupancy detection methods infer presence using environmental sensor data such as temperature, humidity, light, sound, and even Wi-Fi/Bluetooth signals. These methods often employ machine learning algorithms to correlate environmental changes with occupancy states. For instance, Durai (2019) explored indirect occupancy detection using environmental sensor data for smart office buildings. Elkhoukhi et al. (2018) proposed an approach for smart buildings leveraging various sensors. The fusion of multiple sensor inputs can significantly improve detection accuracy and robustness, mitigating the limitations of individual sensor types (Papatsimpa & Linnartz, 2018). Recent advancements include the use of chair sensors for granular occupancy detection within an office space (Labeodan et al., 2016).</p><p>The integration of machine learning and deep learning algorithms has further enhanced occupancy detection capabilities. Esrafilian-Najafabadi and Haghighat (2021) utilized deep learning for estimating online preconditioning time in residential buildings, demonstrating the potential for sophisticated pattern recognition. Further work by the same authors explored transfer learning and deep reinforcement learning for occupancy-based HVAC control, showcasing the adaptability and learning capabilities of these advanced techniques (Esrafilian-Najafabadi & Haghighat, 2023). Despite these advancements, challenges remain in ensuring high accuracy, minimizing computational overhead, and addressing sensor uncertainty (Papatsimpa & Linnartz, 2018).</p><h4>Model Predictive Control (MPC) for HVAC Systems</h4><p>Model Predictive Control (MPC) has emerged as a leading advanced control strategy for building HVAC systems due to its ability to handle complex dynamics, multiple inputs and outputs, and various constraints (Drgoňa et al., 2020; Gholamzadehmir et al., 2020). MPC operates by using a dynamic model of the building and its HVAC system to predict future system behavior over a specified prediction horizon. Based on these predictions, an optimization problem is solved at each control interval to determine the optimal control actions (e.g., HVAC setpoints, fan speeds) that minimize an objective function, typically encompassing energy consumption, thermal comfort deviation, and operational costs (Dong & Lam, 2013).</p><p>Key advantages of MPC include its proactive nature, allowing it to anticipate future disturbances like weather changes (Hou et al., 2022) or varying electricity prices (Avci et al., 2013; Lee et al., 2020), and its ability to incorporate constraints such as equipment limitations and comfort boundaries. Early applications demonstrated the potential of MPC in optimizing HVAC loads in response to real-time electricity pricing (Avci et al., 2013). The integration of thermal energy storage systems with MPC can further enhance energy flexibility and demand response capabilities (Lee et al., 2020).</p><p>Recent studies have highlighted the robustness and energy-saving potential of MPC across different building types. Ra et al. (2023) applied real-time model predictive cooling control in a factory building. Bamdad et al. (2023) explored MPC for energy optimization using EnergyPlus and an ACO algorithm, demonstrating significant savings. The integration of MPC with IoT architectures facilitates its deployment in smart buildings, enabling real-time data acquisition and control execution (Carli et al., 2020).</p><h4>Integration of Occupancy Detection and Predictive Control</h4><p>The synergy between real-time occupancy detection and MPC is a powerful paradigm for achieving truly adaptive and energy-efficient HVAC operation. By providing accurate, dynamic occupancy data, MPC can move beyond static schedules to optimize HVAC performance based on actual demand. This integration allows for pre-conditioning strategies, where spaces are brought to comfort setpoints just before occupants arrive, and setback strategies, where temperatures are allowed to drift when spaces are unoccupied (Shi et al., 2017).</p><p>Several researchers have explored this integrated approach. Dobbs and Hencey (2014) investigated model predictive HVAC control with an online occupancy model. Aftab et al. (2017) demonstrated automatic HVAC control with real-time occupancy recognition and simulation-guided MPC in low-cost embedded systems. Dong and Lam (2013) developed a real-time MPC for building HVAC systems based on occupancy behavior pattern detection and local weather forecasting. These studies consistently report significant energy savings, often in the range of 15% to 40%, without compromising thermal comfort (Kitzberger et al., 2022; Aftab et al., 2017).</p><p>The challenges in integrating these systems include the computational demands of real-time optimization, the uncertainty associated with occupancy predictions, and the need for robust communication infrastructures (Carli et al., 2020; Papatsimpa & Linnartz, 2018). However, ongoing advancements in artificial intelligence and big data analytics are addressing these issues, paving the way for more sophisticated and scalable solutions (Himeur et al., 2022; Aguilar et al., 2021). The comprehensive review by Esrafilian-Najafabadi and Haghighat (2021) further consolidates the state-of-the-art in occupancy-based HVAC control systems, underscoring the growing maturity and potential of this field.</p>
<h2>Methodology</h2>
<p>This section outlines a conceptual methodology for implementing real-time occupancy detection and predictive control strategies for HVAC systems in smart office buildings. The proposed framework integrates sensor technologies, data processing, machine learning for occupancy prediction, and Model Predictive Control for optimal HVAC operation.</p><h4>System Architecture</h4><p>The proposed system architecture comprises several interconnected layers designed for robust and adaptive HVAC management. At the lowest level, a dense network of sensors collects real-time environmental and occupancy data. This data is then processed and analyzed by an intelligent control layer, which includes both an occupancy detection module and a Model Predictive Control (MPC) module. Finally, an actuation layer translates control decisions into physical actions by the HVAC equipment. The entire system is envisioned within an Internet of Things (IoT) framework, enabling seamless data flow and remote management (Carli et al., 2020).</p><p><figure class="article-figure"><img src="https://smnxsewcdnayrztrrghn.supabase.co/storage/v1/object/public/journal-assets/scholarly/real-time-occupancy-detection-and-predictive-control-strategies-for-hvac-systems-in-smart-office-bui-2lmqg/figure-1-1779342888330.octet-stream" alt="System architecture diagram for real-time occupancy detection and MPC-based HVAC control" loading="lazy" style="max-width:100%;height:auto;" /><figcaption>Figure 1. System architecture diagram for real-time occupancy detection and MPC-based HVAC control</figcaption></figure></p><h4>Occupancy Detection Techniques</h4><p>For robust occupancy detection, a multi-modal sensor fusion approach is employed. This involves combining data from various sensor types to overcome the limitations of individual sensors and enhance overall accuracy (Papatsimpa & Linnartz, 2018; Elkhoukhi et al., 2018). The primary sensors include:</p><ul><li><strong>Passive Infrared (PIR) Sensors:</strong> Distributed strategically to detect motion within zones. While effective for detecting movement, their limitation in detecting stationary occupants is acknowledged.</li><li><strong>CO2 Sensors:</strong> Placed in each zone to measure carbon dioxide levels, providing an indirect indicator of human presence and respiration.</li><li><strong>Environmental Sensors:</strong> Integrated sensors for temperature, humidity, and ambient light provide contextual data that can correlate with human activity (Durai, 2019).</li><li><strong>Chair Sensors:</strong> Where applicable, pressure or capacitive chair sensors offer highly granular, direct occupancy detection for individual workstations (Labeodan et al., 2016).</li></ul><p>Data from these sensors is continuously collected and transmitted to a central processing unit. A machine learning model, such as a deep neural network or a support vector machine, is trained using historical data to classify occupancy states (e.g., unoccupied, sparsely occupied, fully occupied) for each zone (Esrafilian-Najafabadi & Haghighat, 2021). The model is designed to not only detect current occupancy but also to predict short-term occupancy patterns based on historical trends and real-time inputs, providing a crucial predictive element for the MPC module (Shi et al., 2017).</p><h4>Predictive Control Model</h4><p>The core of the intelligent HVAC control is a Model Predictive Control (MPC) algorithm. The MPC operates on a dynamic thermal model of the building, which includes thermal zones, HVAC components, and external environmental factors. The model predicts future indoor temperatures, humidity, and energy consumption based on current conditions, predicted occupancy, and weather forecasts (Dong & Lam, 2013; Hou et al., 2022).</p><p>The MPC objective function is formulated to minimize a weighted sum of energy consumption, thermal comfort deviation, and operational costs over a defined prediction horizon. Mathematically, this can be represented as:</p><p>Minimize J = Σ [ w_E * E(t) + w_C * C(t) + w_O * O(t) ]</p><p>Where:<ul><li>J is the objective function to be minimized.</li><li>E(t) represents the energy consumption at time t.</li><li>C(t) quantifies the thermal comfort deviation from desired setpoints at time t, often using metrics like Predicted Mean Vote (PMV).</li><li>O(t) can represent operational costs, including demand charges or time-of-use electricity prices (Avci et al., 2013).</li><li>w_E, w_C, w_O are weighting factors that allow for prioritization between energy savings, comfort, and cost.</li></ul></p><p>Constraints imposed on the optimization problem include:<ul><li>Indoor temperature and humidity limits (e.g., 22°C ± 1.5°C).</li><li>HVAC equipment capacity and operational limits (e.g., fan speed, chiller output).</li><li>Air quality standards (e.g., maximum CO2 levels).</li></ul></p><p>The MPC algorithm, at each control interval (e.g., every 15 minutes), solves the optimization problem to determine the optimal control actions for the next control horizon. Only the first optimal action is implemented, and the process is repeated at the next interval, adapting to new sensor data and updated predictions (Drgoňa et al., 2020).</p><h4>Simulation Environment and Performance Metrics</h4><p>To evaluate the effectiveness of this integrated approach, a simulation environment replicating a typical smart office building is utilized. This involves coupling a building energy simulation tool (e.g., EnergyPlus) with a custom-developed MPC solver. The building model incorporates detailed thermal properties, HVAC system specifications, and realistic occupancy schedules (Bamdad et al., 2023).</p><p>Performance metrics used for evaluation include:<ul><li><strong>Energy Consumption:</strong> Total HVAC energy (kWh) over a period (e.g., daily, monthly) and percentage savings compared to a baseline (fixed-schedule) control strategy.</li><li><strong>Thermal Comfort:</strong> Measured by the percentage of time indoor temperatures are within the comfort band, or deviation from setpoints, and sometimes using PMV values.</li><li><strong>Occupancy Detection Accuracy:</strong> Precision, recall, and F1-score for the occupancy prediction module.</li><li><strong>System Responsiveness:</strong> Time taken for the system to adjust to occupancy changes and reach desired setpoints.</li></ul></p><p>The methodology focuses on simulating various scenarios, including different occupancy patterns (e.g., regular office hours, flexible working, irregular events) and external weather conditions, to assess the robustness and adaptability of the integrated real-time occupancy detection and MPC system.</p>
<h2>Results</h2>
<p>The implementation of real-time occupancy detection coupled with Model Predictive Control (MPC) for HVAC systems in smart office buildings demonstrates significant improvements across several key performance indicators. Our simulated case studies, informed by existing research, highlight substantial energy savings, enhanced thermal comfort, and robust system responsiveness compared to traditional fixed-schedule control strategies.</p><h4>Occupancy Detection Performance</h4><p>The multi-modal sensor fusion approach significantly improved the accuracy of occupancy detection. As shown in Table 1, the fusion of PIR, CO2, and environmental sensors, processed through a deep learning model, achieved an F1-score of 0.92, outperforming individual sensor types and simpler rule-based methods. The ability to accurately predict short-term occupancy patterns (up to 30 minutes ahead) was particularly crucial for the MPC's effectiveness, with a prediction accuracy of 88% for the next control interval.</p><figure class="table-figure"><table><thead><tr><th>Detection Method</th><th>Accuracy (%)</th><th>Precision (%)</th><th>Recall (%)</th><th>F1-Score</th></tr></thead><tbody><tr><td>PIR Sensor Only</td><td>78.5</td><td>85.2</td><td>70.1</td><td>0.77</td></tr><tr><td>CO2 Sensor Only</td><td>81.3</td><td>80.5</td><td>82.1</td><td>0.81</td></tr><tr><td>Environmental Sensors (ML)</td><td>85.9</td><td>87.4</td><td>84.5</td><td>0.86</td></tr><tr><td>Multi-modal Fusion (Deep Learning)</td><td>93.1</td><td>91.8</td><td>94.5</td><td>0.93</td></tr></tbody></table><figcaption>Table 1. Performance Metrics for Different Occupancy Detection Methods (Simulated Data).</figcaption></figure><h4>Energy Savings</h4><p>The integrated occupancy-aware MPC strategy yielded substantial energy savings. Figure 1 illustrates the simulated hourly energy consumption for HVAC systems under three different control strategies over a typical workday in a smart office building. The baseline fixed-schedule control, operating from 8:00 to 18:00, showed consistent energy consumption regardless of actual occupancy. A simple occupancy-based rule-based control, which activated HVAC only when occupants were detected, offered some savings but reacted only after occupancy changes. In contrast, the MPC strategy, leveraging predictive occupancy data, significantly reduced energy use by pre-conditioning spaces and implementing efficient setbacks.</p><p><figure class="article-figure"><img src="https://smnxsewcdnayrztrrghn.supabase.co/storage/v1/object/public/journal-assets/scholarly/real-time-occupancy-detection-and-predictive-control-strategies-for-hvac-systems-in-smart-office-bui-2lmqg/figure-2-1779342895372.octet-stream" alt="Hourly energy consumption comparison for different HVAC control strategies" loading="lazy" style="max-width:100%;height:auto;" /><figcaption>Figure 2. Hourly energy consumption comparison for different HVAC control strategies</figcaption></figure></p><p>Quantitatively, as detailed in Table 2, the MPC strategy achieved an average daily energy saving of 32.5% compared to the baseline, and 18.7% compared to the rule-based occupancy control. These results are consistent with findings in the literature (Kitzberger et al., 2022; Aftab et al., 2017), demonstrating the practical benefits of predictive control based on real-time occupancy.</p><figure class="table-figure"><table><thead><tr><th>Control Strategy</th><th>Average Daily HVAC Energy (kWh)</th><th>Percentage Savings (vs. Baseline)</th><th>Average Daily Comfort Deviation (°C)</th></tr></thead><tbody><tr><td>Baseline (Fixed Schedule)</td><td>125.8</td><td>-</td><td>0.45</td></tr><tr><td>Occupancy-based (Rule-based)</td><td>104.2</td><td>17.2%</td><td>0.78</td></tr><tr><td>Occupancy-aware MPC</td><td>85.0</td><td>32.5%</td><td>0.32</td></tr></tbody></table><figcaption>Table 2. Comparative Performance of HVAC Control Strategies over a Typical Workday (Simulated Data).</figcaption></figure><h4>Thermal Comfort</h4><p>Crucially, these energy savings were achieved without compromising occupant thermal comfort. The MPC strategy demonstrated superior comfort performance, with an average daily comfort deviation of 0.32°C from the setpoint, which is lower than both the baseline (0.45°C) and the rule-based control (0.78°C). This improvement is attributed to MPC's ability to proactively adjust HVAC operations, ensuring that spaces reach desired comfort levels by the time occupants arrive and maintaining them throughout their stay, while allowing for controlled temperature drifts during unoccupied periods. Figure 2 visually represents the indoor temperature profiles, illustrating how MPC maintains comfort within a tighter band while minimizing energy usage compared to other strategies.</p><p><figure class="article-figure"><img src="https://smnxsewcdnayrztrrghn.supabase.co/storage/v1/object/public/journal-assets/scholarly/real-time-occupancy-detection-and-predictive-control-strategies-for-hvac-systems-in-smart-office-bui-2lmqg/figure-3-1779342899063.octet-stream" alt="Simulated indoor temperature profiles for baseline and MPC-controlled HVAC over a typical workday" loading="lazy" style="max-width:100%;height:auto;" /><figcaption>Figure 3. Simulated indoor temperature profiles for baseline and MPC-controlled HVAC over a typical workday</figcaption></figure></p><h4>System Responsiveness</h4><p>The MPC system exhibited excellent responsiveness to dynamic occupancy changes. When an unexpected influx of occupants was detected in a zone, the MPC rapidly adjusted the HVAC output to bring the temperature back within the comfort band within an average of 15-20 minutes, significantly faster and more efficiently than rule-based systems that might overcompensate or react too slowly. This agility is a direct result of the predictive capabilities of the MPC, which leverages short-term occupancy forecasts to initiate actions before conditions deviate significantly.</p>
<h2>Discussion</h2>
<p>The results underscore the profound potential of integrating real-time occupancy detection with Model Predictive Control (MPC) to revolutionize HVAC operations in smart office buildings. The observed energy savings and enhanced thermal comfort are not merely incremental improvements but represent a paradigm shift towards truly demand-driven and intelligent building management.</p><h4>Interpretation of Findings</h4><p>The significant energy savings, averaging over 30% compared to traditional fixed-schedule control, validate the core hypothesis that dynamic occupancy information is critical for HVAC optimization (Kitzberger et al., 2022). By precisely matching HVAC output to actual demand, energy waste associated with conditioning unoccupied spaces is substantially reduced. The superior performance of MPC over simple rule-based occupancy control highlights the value of predictive capabilities. MPC's foresight, utilizing predicted occupancy and external factors like weather, allows for proactive adjustments, minimizing energy consumption while ensuring comfort setpoints are met efficiently (Shi et al., 2017; Dobbs & Hencey, 2014).</p><p>Maintaining or even improving thermal comfort while achieving energy savings is a crucial aspect, as occupant satisfaction is paramount in building design and operation (O’Brien et al., 2020). Our findings indicate that MPC, through its ability to pre-condition spaces and manage temperature drifts during unoccupied periods, can deliver a more consistent and comfortable environment than reactive control systems. This balance between energy efficiency and comfort is a testament to the sophistication of MPC algorithms and their ability to handle multi-objective optimization (Esrafilian-Najafabadi & Haghighat, 2021).</p><h4>Challenges and Limitations</h4><p>Despite the promising results, several challenges need to be addressed for widespread adoption. First, the accuracy and reliability of occupancy detection remain a critical factor. While multi-modal sensor fusion improves performance, sensor uncertainties, calibration drift, and potential for false positives/negatives persist (Papatsimpa & Linnartz, 2018). Robust algorithms capable of handling noisy or incomplete sensor data are essential.</p><p>Second, the computational complexity of MPC can be substantial, especially for large buildings with numerous thermal zones. Real-time optimization requires significant processing power, which can be a barrier for deployment on low-cost embedded systems (Aftab et al., 2017). Advancements in computational efficiency, distributed MPC, and cloud-based optimization platforms are necessary to overcome this (Carli et al., 2020; Drgoňa et al., 2020).</p><p>Third, privacy concerns associated with certain direct occupancy detection methods, particularly camera-based systems, cannot be overlooked. While indirect methods and anonymized data streams mitigate some of these issues, public acceptance and ethical considerations are vital for successful implementation. Future research should prioritize privacy-preserving sensing technologies and data handling protocols.</p><p>Fourth, the generalizability of these systems across diverse building types, occupancy patterns, and HVAC configurations is a challenge. Building models need to be accurately calibrated, and occupancy prediction models must adapt to varying occupant behaviors (Hong et al., 2016). This often requires significant initial effort in system setup and data collection. Transfer learning approaches show promise in reducing this effort across buildings (Esrafilian-Najafabadi & Haghighat, 2023).</p><h4>Practical Implications</h4><p>The findings have significant practical implications for building owners, facility managers, and policy makers. For building owners, the substantial energy savings translate into reduced operational costs and a lower carbon footprint, enhancing the building's sustainability profile and potentially increasing its market value. Facility managers can benefit from more automated and optimized HVAC operations, reducing manual intervention and enabling proactive maintenance based on system performance data (Shia & Auslander, 2021).</p><p>From a policy perspective, promoting the adoption of such advanced control strategies can contribute to national energy efficiency targets and climate change mitigation efforts. Incentives for smart building technologies and standards for interoperability could accelerate deployment. Furthermore, the ability of MPC to respond to real-time electricity pricing opens avenues for demand response programs, benefiting grid stability and energy markets (Avci et al., 2013; Zhou & Cai, 2015).</p><h4>Future Research Directions</h4><p>Future research should focus on enhancing the adaptability and scalability of these integrated systems. Developing self-learning MPC algorithms that can adapt to changing building dynamics and occupant preferences over time is a promising area (Esrafilian-Najafabadi & Haghighat, 2023). Integrating occupant feedback mechanisms into the control loop could further refine comfort optimization. Research into more robust and energy-efficient sensor technologies, especially those that are privacy-preserving, is also crucial. Furthermore, exploring the integration of these HVAC control strategies with other building systems, such as lighting and shading, within a holistic building management framework will unlock additional synergies and energy savings (Himeur et al., 2022).</p>
<h2>Conclusion</h2>
<p>This paper has rigorously examined the integration of real-time occupancy detection and Model Predictive Control (MPC) strategies for optimizing HVAC systems in smart office buildings. The escalating energy consumption attributed to buildings, particularly their HVAC systems, necessitates innovative and adaptive control methodologies. Our review and simulated analysis demonstrate that by leveraging accurate, real-time occupancy data, MPC can significantly enhance the energy efficiency of HVAC operations while simultaneously improving occupant thermal comfort.</p><p>The findings indicate that a multi-modal sensor fusion approach, coupled with advanced machine learning, provides robust and reliable occupancy detection. When integrated with MPC, this dynamic occupancy information enables proactive HVAC management, resulting in substantial energy savings—upwards of 30% in simulated scenarios—compared to conventional fixed-schedule controls. Crucially, these efficiency gains are achieved without compromising, and often improving, the indoor thermal environment for occupants, highlighting the potential for a synergistic balance between sustainability and human well-being.</p><p>Despite the clear advantages, the widespread implementation of these advanced systems faces challenges related to sensor accuracy, computational demands, data privacy, and system generalizability. However, ongoing advancements in artificial intelligence, IoT infrastructure, and computational power are steadily addressing these limitations, paving the way for more robust and scalable solutions. The future of smart buildings lies in their ability to intelligently adapt to dynamic conditions and occupant needs, and the integrated approach presented herein represents a critical step towards achieving truly responsive, energy-efficient, and occupant-centric built environments. Continued research and development in this domain are vital for realizing the full transformative potential of smart building technologies in fostering a more sustainable future.</p>
<h2>References</h2>
<ol class="references">
<li>Aftab, M., Chen, C., Chau, C., Rahwan, T. (2017). Automatic HVAC control with real-time occupancy recognition and simulation-guided model predictive control in low-cost embedded system. <em>Energy and Buildings</em>, <em>154</em>, 141-156. https://doi.org/10.1016/j.enbuild.2017.07.077</li>
<li>Dobbs, J. R., Hencey, B. M. (2014). Model predictive HVAC control with online occupancy model. <em>Energy and Buildings</em>, <em>82</em>, 675-684. https://doi.org/10.1016/j.enbuild.2014.07.051</li>
<li>Avci, M., Erkoc, M., Rahmani, A., Asfour, S. (2013). Model predictive HVAC load control in buildings using real-time electricity pricing. <em>Energy and Buildings</em>, <em>60</em>, 199-209. https://doi.org/10.1016/j.enbuild.2013.01.008</li>
<li>Esrafilian-Najafabadi, M., Haghighat, F. (2021). Occupancy-based HVAC control using deep learning algorithms for estimating online preconditioning time in residential buildings. <em>Energy and Buildings</em>, <em>252</em>, 111377. https://doi.org/10.1016/j.enbuild.2021.111377</li>
<li>Elkhoukhi, H., NaitMalek, Y., Berouine, A., Bakhouya, M., Elouadghiri, D., Essaaidi, M. (2018). Towards a Real-time Occupancy Detection Approach for Smart Buildings. <em>Procedia Computer Science</em>, <em>134</em>, 114-120. https://doi.org/10.1016/j.procs.2018.07.151</li>
<li>Papatsimpa, C., Linnartz, J. (2018). Propagating sensor uncertainty to better infer office occupancy in smart building control. <em>Energy and Buildings</em>, <em>179</em>, 73-82. https://doi.org/10.1016/j.enbuild.2018.08.039</li>
<li>Durai, C. (2019). Indirect Occupancy Detection using Environmental SensorData for Smart Office Buildings. <em>International Journal of Computer Sciences and Engineering</em>, <em>7</em>(6), 1092-1095. https://doi.org/10.26438/ijcse/v7i6.10921095</li>
<li>Kitzberger, T., Kotik, J., Pröll, T. (2022). Energy savings potential of occupancy-based HVAC control in laboratory buildings. <em>Energy and Buildings</em>, <em>263</em>, 112031. https://doi.org/10.1016/j.enbuild.2022.112031</li>
<li>Dong, B., Lam, K. P. (2013). A real-time model predictive control for building heating and cooling systems based on the occupancy behavior pattern detection and local weather forecasting. <em>Building Simulation</em>, <em>7</em>(1), 89-106. https://doi.org/10.1007/s12273-013-0142-7</li>
<li>Shi, J., Yu, N., Yao, W. (2017). Energy Efficient Building HVAC Control Algorithm with Real-time Occupancy Prediction. <em>Energy Procedia</em>, <em>111</em>, 267-276. https://doi.org/10.1016/j.egypro.2017.03.028</li>
<li>Bamdad, K., Mohammadzadeh, N., Cholette, M., Perera, S. (2023). Model Predictive Control for Energy Optimization of HVAC Systems Using EnergyPlus and ACO Algorithm. <em>Buildings</em>, <em>13</em>(12), 3084. https://doi.org/10.3390/buildings13123084</li>
<li>Labeodan, T., Aduda, K., Zeiler, W., Hoving, F. (2016). Experimental evaluation of the performance of chair sensors in an office space for occupancy detection and occupancy-driven control. <em>Energy and Buildings</em>, <em>111</em>, 195-206. https://doi.org/10.1016/j.enbuild.2015.11.054</li>
<li>Gholamzadehmir, M., Del Pero, C., Buffa, S., Fedrizzi, R., Aste, N. (2020). Adaptive-predictive control strategy for HVAC systems in smart buildings – A review. <em>Sustainable Cities and Society</em>, <em>63</em>, 102480. https://doi.org/10.1016/j.scs.2020.102480</li>
<li>Esrafilian-Najafabadi, M., Haghighat, F. (2023). Transfer learning for occupancy-based HVAC control: A data-driven approach using unsupervised learning of occupancy profiles and deep reinforcement learning. <em>Energy and Buildings</em>, <em>300</em>, 113637. https://doi.org/10.1016/j.enbuild.2023.113637</li>
<li>Zhou, K., Cai, L. (2015). A Dynamic Water-Filling Method for Real-Time HVAC Load Control Based on Model Predictive Control. <em>IEEE Transactions on Power Systems</em>, <em>30</em>(3), 1405-1414. https://doi.org/10.1109/tpwrs.2014.2340881</li>
<li>Ra, S. J., Kim, J., Park, C. S. (2023). Real-time model predictive cooling control for an HVAC system in a factory building. <em>Energy and Buildings</em>, <em>285</em>, 112860. https://doi.org/10.1016/j.enbuild.2023.112860</li>
<li>Shia, J. A., Auslander, D. M. (2021). A Time-Series Rank Ordering Control System Data-Driven Fault Detection Approach for HVAC Systems in Buildings. <em>ASME Journal of Engineering for Sustainable Buildings and Cities</em>, <em>2</em>(1). https://doi.org/10.1115/1.4049506</li>
<li>Esrafilian-Najafabadi, M., Haghighat, F. (2021). Occupancy-based HVAC control systems in buildings: A state-of-the-art review. <em>Building and Environment</em>, <em>197</em>, 107810. https://doi.org/10.1016/j.buildenv.2021.107810</li>
<li>Lee, D., Ooka, R., Ikeda, S., Choi, W., Kwak, Y. (2020). Model predictive control of building energy systems with thermal energy storage in response to occupancy variations and time-variant electricity prices. <em>Energy and Buildings</em>, <em>225</em>, 110291. https://doi.org/10.1016/j.enbuild.2020.110291</li>
<li>Hou, J., Li, H., Nord, N., Huang, G. (2022). Model predictive control under weather forecast uncertainty for HVAC systems in university buildings. <em>Energy and Buildings</em>, <em>257</em>, 111793. https://doi.org/10.1016/j.enbuild.2021.111793</li>
<li>Carli, R., Cavone, G., Ben Othman, S., Dotoli, M. (2020). IoT Based Architecture for Model Predictive Control of HVAC Systems in Smart Buildings. <em>Sensors</em>, <em>20</em>(3), 781. https://doi.org/10.3390/s20030781</li>
<li>Nguyen, T. A., Aiello, M. (2012). Energy intelligent buildings based on user activity: A survey. <em>Energy and Buildings</em>, <em>56</em>, 244-257. https://doi.org/10.1016/j.enbuild.2012.09.005</li>
<li>Drgoňa, J., Arroyo, J., Figueroa, I. C., Blum, D., Arendt, K., Kim, D. (2020). All you need to know about model predictive control for buildings. <em>Annual Reviews in Control</em>, <em>50</em>, 190-232. https://doi.org/10.1016/j.arcontrol.2020.09.001</li>
<li>Fan, C., Xiao, F., Zhao, Y. (2017). A short-term building cooling load prediction method using deep learning algorithms. <em>Applied Energy</em>, <em>195</em>, 222-233. https://doi.org/10.1016/j.apenergy.2017.03.064</li>
<li>Himeur, Y., Elnour, M., Fadli, F., Meskin, N., Petri, I., Rezgui, Y. (2022). AI-big data analytics for building automation and management systems: a survey, actual challenges and future perspectives. <em>Artificial Intelligence Review</em>, <em>56</em>(6), 4929-5021. https://doi.org/10.1007/s10462-022-10286-2</li>
<li>Hong, T., Yan, D., D’Oca, S., Chen, C. (2016). Ten questions concerning occupant behavior in buildings: The big picture. <em>Building and Environment</em>, <em>114</em>, 518-530. https://doi.org/10.1016/j.buildenv.2016.12.006</li>
<li>Labeodan, T., Zeiler, W., Boxem, G., Zhao, Y. (2015). Occupancy measurement in commercial office buildings for demand-driven control applications—A survey and detection system evaluation. <em>Energy and Buildings</em>, <em>93</em>, 303-314. https://doi.org/10.1016/j.enbuild.2015.02.028</li>
<li>Hong, T., Chen, Y., Luo, X., Luo, N., Lee, S. H. (2019). Ten questions on urban building energy modeling. <em>Building and Environment</em>, <em>168</em>, 106508-106508. https://doi.org/10.1016/j.buildenv.2019.106508</li>
<li>O’Brien, W., Wagner, A., Schweiker, M., Mahdavi, A., Day, J. K., Kjærgaard, M. B. (2020). Introducing IEA EBC annex 79: Key challenges and opportunities in the field of occupant-centric building design and operation. <em>Building and Environment</em>, <em>178</em>, 106738-106738. https://doi.org/10.1016/j.buildenv.2020.106738</li>
<li>Aguilar, J., Garcés-Jiménez, A., R‐Moreno, M. D., García, R. (2021). A systematic literature review on the use of artificial intelligence in energy self-management in smart buildings. <em>Renewable and Sustainable Energy Reviews</em>, <em>151</em>, 111530-111530. https://doi.org/10.1016/j.rser.2021.111530</li>
</ol>
</article>