Previous Article in Journal
Planning Resources by Model Predictive Control
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Engineering Proceedings
Volume 100
Issue 1
10.3390/engproc2025100019
engproc-logo
Submit to this Journal
► Article Menu

Article Menu

share Share announcement Help format_quote Cite
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Proceeding Paper

Applications of Virtual Reality Simulations and Machine Learning Algorithms in High-Risk Environments †

by
Velyo Vasilev
*,
Dilyana Budakova
and
Veselka Petrova-Dimitrova
Faculty of Electronics and Automation, Technical University of Sofia, Branch Plovdiv, Tsanko Dyustabanov 25, 4000 Plovdiv, Bulgaria
*
Author to whom correspondence should be addressed.
†
Presented at the 14th International Scientific Conference TechSys 2025—Engineering, Technology and Systems, Plovdiv, Bulgaria, 15–17 May 2025.
Eng. Proc. 2025, 100(1), 19; https://doi.org/10.3390/engproc2025100019 (registering DOI)
Published: 7 July 2025
(This article belongs to the Proceedings of The 14th International Scientific Conference TechSys 2025—Engineering, Technologies and Systems)
Versions Notes

Abstract

In this article, the application of virtual reality technology for the realistic and immersive visualization of various tasks and scenarios in fields such as power engineering and fire safety has been examined in order to help prepare students and professional electrical engineers with electrical safety, the operation of electrical substations, potential emergencies, injury prevention, fire safety, and others. Additionally, the use of machine learning algorithms to guide evacuations from hazardous environments, fault prevention, fire prediction, and discovery of conductive materials has been examined. The most frequently used algorithms in these areas have also been described and summarized, and conclusions have been made about the combined advantages of using VR and ML algorithms. Finally, the needs, contributions, and challenges of using machine learning in virtual reality projects have been examined.

1. Introduction

Virtual reality is a subfield of computer vision. It is closely related to areas such as computer graphics, computer animation, or human–computer interaction [1,2]. In this technology, the environment is realized entirely by computer graphics and is completely digital. Building a realistic 3D image of real-world scenes and the ability to interact with virtual objects in real time require large computational resources and data, sensors, HMD—head-mounted display—and joysticks for a highly immersive experience. The architecture of the virtual reality system is given in [3].
Combining the realistic immersive potential of virtual reality and the analytical capabilities of machine learning leads to the creation of highly engaging, strictly individual, adaptive, completely safe, and professionally implemented training systems [4]. These technologies can be used to create 3D models of buildings, interactive and engaging advertisements presented in a unique way, and realistic training in a safe environment for pilots, surgeons, soldiers, and electrical engineers. Machine learning is a current solution for improving the speed and automation of image processing in virtual and augmented reality applications [5]. The integration of machine learning leads to better quality of the rendered image in real time [6]. Machine learning algorithms must be adapted for real-time use to provide a response to user interaction with the virtual environment. Tools for automatic generation of datasets are applied. Trained deep neural networks use this data to detect and track objects and markers in augmented reality applications and for real-time virtual reality [5]. The huge dataset required to train deep neural networks can reduce the efficiency and accuracy of VR applications [7]. Deep reinforcement learning algorithms are one solution to avoid the problem of requiring a lot of labeled training data. They use trial and error and models for their training. One solution is to use techniques such as in-stream supervision [8], which automate the labeling of data during its use. Another solution is data transfer, which can be implemented through the machine learning technique transfer learning [9]. In this machine learning technique, knowledge from trained models and datasets for one task is used to improve the performance of another similar task to be solved. Machine learning models need to include optimization techniques to be able to adapt to dynamic data. In [10], an optimized Bidirectional Long Short-Term Memory (Bi-LSTM) model enhanced with Firefly Optimization (FFO) is proposed. Machine learning allows for personalization of scenarios implemented in virtual reality, which is based on data and real-time adaptation. In this way, patients and trained professionals can predict their specific indicators, such as motor activity, cognitive engagement, and physiological responses, and can optimize the tasks for performance. Dynamic adaptation of the training is proposed so that it remains challenging and achievable for each user [10,11].
Virtual reality technology, supported by machine learning, natural language processing, and robotics, is used to build real-world representations based on large-scale models. They are used together for medical training for diagnosis and treatment [12], for attention prediction [13], for stroke rehabilitation [10,14], for designing and generating virtual architecture of buildings supported by deep learning and using datasets [15], for hazardous environments without exposing people and equipment to risk in the energy sector [16,17,18], and for human–robot interaction in manufacturing [19,20]. And last but not least, projects that combine virtual reality, virtual twin, and digital twin technologies will lead to new, even larger possibilities and results [21].
This article is organized as follows: First, several virtual reality simulations are discussed. Their models, uses, and areas of application have been analyzed and summarized in a table. The second point discusses machine learning algorithms and classifies them based on their usage in certain cases of work to ensure safety in high-risk environments. In the third section, the algorithms that find uses in all of the examined fields are researched and described. The last section presents a discussion.

2. Uses of Virtual Reality for Safety and Education

Several applications for simulation using virtual and augmented reality have been developed [22,23,24]. This technology facilitates realistic learning in a safe environment [25,26], as demonstrated by a 3D model of an electrical substation [27]. Models of switch rooms, server rooms, and data centers have been developed. Additionally, there are detailed recreations of substations and their equipment, facilitating distance learning for substation operators and other professionals [28].
An electrostatic platform has been created for collaborative, simulation-based exploratory training in immersive virtual reality in reference [29]. Providing innovative training approaches through interactive 3D virtual reality lessons is of interest to several modern technology companies in the power system. The lessons cover training on maintaining workplace safety by complying with Occupational Safety and Health Administration (OSHA) standards, transformer oil sampling, familiarization with high-voltage electrical substation equipment such as power transformers, oil circuit breakers, close devices, switchgear, reviewing 3D models of the constituent components of the equipment, a training system for power line operators, etc. [30]. Another example of immersive virtual training for substation electricians is [31,32]. In Table 1, a compilation of the models used, their practical application, and the area in which the models are set to train participants has been made.

3. Uses of Machine Learning Algorithms in High-Risk Environments

Four different applications of machine learning algorithms related to hazardous environments have been examined in the following order: evacuation from a dangerous environment, electrical equipment fault prevention, fire forecast and prevention, and discovery of conductive materials.
Shown in Table 2 is a collection of algorithms used in the aforementioned fields. The first column describes the areas of use in which machine learning has been applied. The second column lists algorithms, networks, and functions that are known to be used to solve problems in those fields. The four sectors and the ML techniques contributing to them are described in a numbered list below.
  • In evacuation scenarios, Deep Q-Networks (DQN) and Q-Networks algorithms are used to optimize evacuation routes and improve the efficiency of evacuation procedures. Other reinforcement learning algorithms, such as Proximal Policy Optimization and Soft Actor-Critic, enable agents to learn from interactions with the environment, allowing them to devise strategies that minimize evacuation time while accounting for the dynamic nature of the environment’s behavior. Imitation learning algorithms such as behavioral cloning (BC) and generative adversarial imitation learning (GAIL) can also find escape routes from dangerous environments, with varying success depending on the number of demonstrations and the dynamic nature of the environment;
  • A wide range of algorithms are used to prevent accidents in industrial settings, including Hidden Markov Models (HMM), Multilayer Perceptron (MLP), and Ant Colony Optimization. Advanced and modified models such as BOA–NARX and ARIMAX are used for time series forecasting, while Linear Dimension Reduction, Maximal Marginal Likelihood Estimation, and Multiclass Support Vector Machines (SVM) help in analyzing complex datasets. Collectively, these algorithms contribute to early detection and prevention of faults and reduce the risks associated with equipment failures.
  • Fire prediction and prevention strategies can be made as a combination of reinforcement learning and traditional prediction models. Algorithms such as value iteration, policy iteration, and Monte Carlo Tree Search simulate and evaluate potential fire spread scenarios. Asynchronous Advantage Actor-Critic (A3C) improves decision-making processes in dynamic environments. Frequency Ratio Multilayer Perceptron (FR-MLP) and linear regression are used to analyze data and identify fire risk factors for timely intervention.
  • The search for new conductive elements is performed using supervised learning techniques, including logistic regression and logistic functions. They can help researchers identify patterns and relationships within experimental data to develop innovative materials with improved conductive properties.

4. Research of the Algorithms with a Wide Range of Application

As is shown in the third column of Table 2, some of the algorithms are used in all four cases—both for material discovery and for operation in various high-risk environments. Support Vector Machines (SVM), Q-learning, neural/artificial networks, decision trees, radial basis functions, support vector regression (SVR), Random Forest, and deep learning techniques have shown effectiveness in various safety, hazard avoidance, and accident prevention applications due to their ability to handle diverse datasets and complex relationships. Below is a description of each of these algorithms and the particular use case:
Support Vector Machine (SVM) is a machine learning method used for both classification and regression by finding the optimal boundary or hyperplane that best separates the classes, called kernel trickery. SVM includes a penalty parameter that balances the width of the field and the classification errors. In [34,35,37], SVM has been applied to predicting forest fire susceptibility by classifying areas as fire-prone based on environmental data, to predicting the maximum electrical output of a baseload combined-cycle power plant, and in the development of smart power systems.
The random forest algorithm is a machine learning method that is applied to classification and regression problems. It creates multiple decision trees using random subsets of data and training functions. Each tree is trained on a different part of the data, and the prediction is made by averaging the results (in regression) or by majority voting (in classification). The accuracy of the algorithm is increased by combining the predictions from different trees, making it more robust to noise and anomalies. Random Forest is widely used because of its ability to handle large amounts of data and complex relationships between individual variables [35,37,41,43,44,45].
The radial basis function (RBF) is used for classification, regression, and interpolation tasks. It measures the similarity between data points based on their distance from a central point. The RBF is typically used as an activation function in the hidden layer, allowing the network to learn nonlinear relationships. The model is typically trained by adjusting the center and spread of the RBF to best fit the data. RBF networks are known for their ability to efficiently process complex and high-dimensional data [36,39].
Deep learning is a subset of machine learning that uses artificial neural networks, built from multiple layers, to model complex patterns from data. During training, deep learning models use algorithms such as backpropagation to minimize the errors between predicted and actual results [46].
Q-learning is a model-free reinforcement learning algorithm that allows an agent to learn optimal behavior in a given environment by exploring different actions and learning from their outcomes. Its basic concept is to determine the best policy for each situation by using a Q-table that stores predictions of the expected future rewards for each action-state pair. Q-learning is widely applied in areas such as robotics, autonomous systems, control systems, and game development, where agents need to make intelligent decisions under changing conditions [34,38].
A decision tree is a machine learning algorithm that makes predictions by dividing data into branches based on feature values, forming a tree-like structure. Each internal node represents a decision based on a function, while the leaves represent the final result or prediction. The decision tree algorithm is used for both classification and regression tasks, enabling predictions and insights by modeling decision-making processes based on input features [34,41,42].
A neural network, or artificial neural network (ANN), is a computational model that mimics the structure and functioning of the human brain. It is built from multiple layers of interconnected nodes, where each node processes information and passes it on to the next layer, learning by adjusting the weights of the connections between neurons during training, often using algorithms such as backpropagation. Neural networks are particularly effective for tasks such as predictive analytics due to their ability to model complex, nonlinear relationships in large datasets [34,36,39,41].

5. Discussion

In virtual reality technology, the goal is to create a realistic and immersive digital environment for users. The focus is on the visualization of real or imagined worlds, user interaction, and user experience. In machine learning, the goal is to implement and use data-learning methods and algorithms for prediction and decision-making without manually programming each task. The focus is on data analysis, prediction, and automated data processing.
Machine learning can be used in virtual reality to personalize content based on user reactions, improve the simulation by predicting user actions, recognize gestures, voice commands, emotions, cognitive perceptions and reactions of users, user states, and the level of difficulty of tasks, and provide feedback to users.
To implement virtual reality with machine learning integration, the following tools and frameworks are required:
  • Unity or Unreal Engine v2023.2.20f1 (for VR development).
  • Python with TensorFlow or PyTorch v2.2.1 (for training ML models).
  • ONNX v1.14.0 or TensorFlow Lite (for integrating trained models into real-time VR apps).
  • Sensor and input integration (Leap Motion, eye trackers, VR controllers).
  • NVIDIA DeepStream v6.0 (for complex video/gesture processing).
  • OpenCV v4.8.1 (for real-time image and video processing).
An example workflow for such a system is as follows: Data collection. Training a machine learning model. Exporting the model. Integrating the model into a virtual reality application. Testing and optimization.
Creating virtual reality (VR) simulations such as virtual electrical substations, fire simulators, and machine learning algorithms can improve the approach to education, prevention, and response in high-risk environments, as simulation technologies offer innovative solutions that enhance safety and preparedness in situations ranging from industrial accidents to natural disasters. Through the use of VR, immersive learning environments can be created that simulate real-world emergencies, allowing beginners and professionals to practice response strategies in a controlled environment. Such experiential learning contributes to better retention of critical skills and improves decision-making under pressure.
Machine learning algorithms play a vital role in analyzing vast amounts of data to discover new elements, predict potential hazards, optimize emergency response strategies, detect fires, prevent damage, and guide evacuations through pathfinding. By processing information in real time, ML algorithms can help identify patterns that may indicate an imminent threat, allowing for proactive measures to be taken. The combination of VR and machine learning not only improves training efficiency but also helps in developing predictive models that improve situational awareness during emergencies.
Perhaps the future lies in the combined use of the capabilities of various technologies such as virtual reality, machine learning, generative artificial intelligence, natural language understanding, gestures, voice commands, virtual twins, and digital twins.

6. Conclusions

The research area of high-risk environments is rapidly evolving, with an increasing number of projects being implemented to improve safety and efficiency. As the demand for effective risk management grows, technologies such as virtual reality (VR) simulators and machine learning (ML) algorithms are becoming indispensable. These innovations are critical in various sectors, including emergency response, hazardous materials handling, evacuation, and risk prevention, where informed decision-making is key. VR simulators provide an opportunity to develop critical skills without the risk of real-world consequences by simulating disaster scenarios to improve responses or rehearse procedures for the safe handling of hazardous materials. ML algorithms are transforming data collection and analysis in high-risk environments. By leveraging predictive analytics and machine learning capabilities, these algorithms can identify patterns and anomalies in operational data, which helps to anticipate potential hazards and optimize resource allocation. They help in real-time monitoring of conditions, predicting equipment failures, and improving decision-making processes.

Author Contributions

Conceptualization, V.V., D.B. and V.P.-D.; methodology, V.V., D.B. and V.P.-D.; software, V.V., D.B. and V.P.-D.; validation, V.V., D.B. and V.P.-D.; formal analysis, V.V., D.B. and V.P.-D.; investigation, V.V., D.B. and V.P.-D.; resources, V.V., D.B. and V.P.-D.; data curation, V.V., D.B. and V.P.-D.; writing—original draft preparation, V.V., D.B. and V.P.-D.; writing—review, and editing, V.V., D.B. and V.P.-D.; visualization, V.V., D.B. and V.P.-D.; supervision, V.V., D.B. and V.P.-D.; project administration, V.V., D.B. and V.P.-D.; funding acquisition, V.V., D.B. and V.P.-D. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding author(s).

Acknowledgments

This research is supported by the Bulgarian Ministry of Education and Science under the second stage of the National Program “Young Scientists and Postdoctoral Students–2”). The authors would like to thank the Research and Development Sector at the Technical University of Sofia for financial support.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
VRVirtual Reality
MLMachine Learning
DQNDeep Q-Network
SACSoft Actor-Critic
PPOProximal Policy Optimization
ILImitation Learning
RLReinforcement Learning
GAILGenerative Adversarial Imitation Learning
BCBehavioral Cloning
SVMSupport Vector Machine
HMMHidden Markov Model
NARXNonlinear autoregressive with external input
SVRSupport Vector Regressor
BOABayesian optimization algorithm
ARIMAXAutoregressive integrated moving average with exogenous inputs
DLDeep Learning

References

  1. Gutiérrez A., M.A.; Vexo, F.; Thalmann, D. Stepping into Virtual Reality; Springer Nature Switzerland: Cham, Switzerland, 2023. [Google Scholar] [CrossRef]
  2. Virtual Reality System—An Overview|ScienceDirect Topics. Available online: https://www.sciencedirect.com/topics/computer-science/virtual-reality-system (accessed on 6 May 2025).
  3. Architecture of Virtual Reality System-Twin Reality. Available online: https://twinreality.in/architecture-of-virtual-reality-system/ (accessed on 6 May 2025).
  4. Sindu, I.G.P.; Hartati, R.S.; Sudarma, M.; Gunantara, N. Systematic literature review of machine learning in virtual reality and augmented reality. J. Nas. Pendidik. Tek. Inform. JANAPATI 2023, 12, 108–118. [Google Scholar] [CrossRef]
  5. Le, H.; Nguyen, M.; Yan, W.Q.; Nguyen, H. Augmented Reality and Machine Learning Incorporation Using YOLOv3 and ARKit. Appl. Sci. 2021, 11, 6006. [Google Scholar] [CrossRef]
  6. Jiang, H.; Padebettu, R.R.; Sakamoto, K.; Bastani, B. Architecture of Integrated Machine Learning in Low Latency Mobile VR Graphics Pipeline. In SIGGRAPH Asia 2019 Technical Briefs, in SA ‘19; Association for Computing Machinery: New York, NY, USA, 2019; pp. 41–44. [Google Scholar] [CrossRef]
  7. FileCloud, T. The Limitations of Machine Learning in an Enterprise Setting. FileCloud blog. Available online: https://www.filecloud.com/blog/2018/06/top-5-limitations-of-machine-learning-in-an-enterprise-setting/ (accessed on 5 May 2025).
  8. Li, Y.; Wang, Y.; Liu, Q.; Bi, C.; Jiang, X.; Sun, S. Incremental semi-supervised learning on streaming data. Pattern Recognit. 2019, 88, 383–396. [Google Scholar] [CrossRef]
  9. What Is Transfer Learning?|IBM. Available online: https://www.ibm.com/think/topics/transfer-learning (accessed on 5 May 2025).
  10. Alsheikhy, A.A.; Shawly, T.; Said, Y.E.; Ahmed, H.E.; Alazzam, M.B. Developing machine learning models for personalized game-based stroke rehabilitation therapy in virtual reality. Alex. Eng. J. 2025, 121, 358–369. [Google Scholar] [CrossRef]
  11. Herne, R.; Shiratuddin, M.F.; Rai, S.; Blacker, D.; Laga, H. Improving Engagement of Stroke Survivors Using Desktop Virtual Reality-Based Serious Games for Upper Limb Rehabilitation: A Multiple Case Study. IEEE Access 2022, 10, 46354–46371. [Google Scholar] [CrossRef]
  12. Daş, F.; Elmas, E.T.; Bucak, İ.Ö. Innovative Use of Machine Learning-Aided Virtual Reality and Natural Language Processing Technologies in Dyslexia Diagnosis and Treatment Phases. In Digital Frontiers-Healthcare, Education, and Society in the Metaverse Era; IntechOpen: Rijeka, Croatia, 2024. [Google Scholar] [CrossRef]
  13. Fathy, F.; Mansour, Y.; Sabry, H.; Refat, M.; Wagdy, A. Virtual reality and machine learning for predicting visual attention in a daylit exhibition space: A proof of concept. Ain Shams Eng. J. 2023, 14, 102098. [Google Scholar] [CrossRef]
  14. Covaciu, F.; Pisla, A.; Iordan, A.-E. Development of a Virtual Reality Simulator for an Intelligent Robotic System Used in Ankle Rehabilitation. Sensors 2021, 21, 1537. [Google Scholar] [CrossRef]
  15. Towards AI-Architecture Liberty: A Comprehensive Survey on Design and Generation of Virtual Architecture by Deep Learning. Available online: https://arxiv.org/html/2305.00510v4 (accessed on 6 May 2025).
  16. Galvan-Bobadilla, I.; Ayala-Garcia, A.; Rodriguez-Gallegos, E.; Arroyo-Figueroa, G. Virtual reality training system for the maintenance of underground lines in power distribution system. In Proceedings of the Third International Conference on Innovative Computing Technology (INTECH 2013), London, UK, 29–31 August 2013; IEEE: London, UK, 2013; pp. 199–204. [Google Scholar] [CrossRef]
  17. Giraldo, J.S.; Kok, K.; Paterakis, N.G. A structured review on game-based learning applied to electrical power and energy engineering. Comp Applic Eng. 2024, 32, e22686. [Google Scholar] [CrossRef]
  18. Lopez, J.M.G.; Betancourt, R.O.J.; Arredondo, J.M.R.; Laureano, E.V.; Haro, F.R. Incorporating Virtual Reality into the Teaching and Training of Grid-Tie Photovoltaic Power Plants Design. Appl. Sci. 2019, 9, 4480. [Google Scholar] [CrossRef]
  19. Loch, F.; Koltun, G.; Karaseva, V.; Pantförder, D.; Vogel-Heuser, B. Model-based training of manual procedures in automated production systems. Mechatronics 2018, 55, 212–223. [Google Scholar] [CrossRef]
  20. Andaluz, V.H.; Pérez, J.A.; Carvajal, C.P.; Ortiz, J.S. Virtual environment for teaching and learning robotics applied to industrial processes. In Augmented Reality, Virtual Reality, and Computer Graphics, Proceedings of the 6th International Conference, AVR 2019, Santa Maria al Bagno, Italy, 24–27 June 2019; Proceedings, Part II 6; Springer International Publishing: Cham, Switzerland, 2019; pp. 442–455. [Google Scholar] [CrossRef]
  21. Fuller, A.; Fan, Z.; Day, C.; Barlow, C. Digital Twin: Enabling Technologies, Challenges and Open Research. IEEE Access 2020, 8, 108952–108971. [Google Scholar] [CrossRef]
  22. Ilie, R.; Shaffer, E.; D’Angelo, C.M.; Cermak, D.; Lin, M.Y.; Chen, H. Virtual reality laboratory experiences for electricity and magnetism courses. In Proceedings of the 2021 ASEE Virtual Annual Conference Content Access, Virtual, 2021 July 26. [Google Scholar]
  23. Pérez, S.S.; Lopez, J.M.G.; Barba, M.A.V.; Betancou, R.O.J.; Solís, J.M.; Ornelas, J.L.R.; García, G.I.R.; Haro, F.R. On the Use of Augmented Reality to Reinforce the Learning of Power Electronics for Beginners. Electronics 2022, 11, 302. [Google Scholar] [CrossRef]
  24. Morelot, S.; Garrigou, A.; Dedieu, J.; NKaoua, B. Virtual reality for fire safety training: Influence of immersion and sense of presence on conceptual and procedural acquisition. Comput. Educ. 2021, 166, 104145. [Google Scholar] [CrossRef]
  25. Digital Engineering and Magic, VR Electrical Safety Training/OSHA VR Training. Available online: https://www.youtube.com/watch?v=D4LarinRBFA (accessed on 10 May 2025).
  26. AAT Training hub, Training For Electrical Safety. Available online: https://aattraininghub.com/services/vr-developersingapore/technology-vr-ar-immersive-singapore-5/ (accessed on 10 May 2025).
  27. Lima, G.F.M.; Cardoso, A.; Lamounier, E.A., Jr.; Macedo, J.R., Jr. Development of a Virtual Reality Environment of Electric Power Substations for Quality of Service Improvement. Available online: https://www.academia.edu/47180297/Development_of_a_Virtual_Reality_environment_of_electric_power_substations_for_quality_of_service_improvement (accessed on 10 May 2025).
  28. E-Spaces, Electrical Substation VR Simulator-E-Spaces. Available online: https://e-spaces.com/electrical-substation-vr-simulator/ (accessed on 10 May 2025).
  29. Greenwald, S.W.; Corning, W.; McDowell, G.; Maes, P.; Belcher, J. ElectroVR: An Electrostatic Playground for Collaborative, Simulation-Based Exploratory Learning in Immersive Virtual Reality. In Proceedings of the International Conference on CSCL 2019 Proceedings, Lyon, France, 17–21 June 2019; pp. 997–1000. [Google Scholar]
  30. Perez-Ramirez, M.; Arroyo-Figueroa, G.; Ayala, A. The use of a virtual reality training system to improve technical skill in the maintenance of live-line power distribution networks. J. Interact. Learn. Environ. 2021, 29, 1–18. [Google Scholar] [CrossRef]
  31. Silva, A.C.; Cardoso, A.; Lamounier, E.A.; Barreto, C.L. Virtual Reality for Monitor and Control of Electrical Substations. An. Acad. Bras. Ciências 2021, 93, e20200267. [Google Scholar] [CrossRef]
  32. Memik, N. The Virtual reality electrical substation field trip: Exploring student perceptions and cognitive learning. STEM Educ. 2021, 1, 47–59. [Google Scholar] [CrossRef]
  33. Xue, Y.; Wu, R.; Liu, J.; Tang, X. Crowd Evacuation Guidance Based on Combined Action Reinforcement Learning. Algorithms 2021, 14, 26. [Google Scholar] [CrossRef]
  34. Ibrahim Sohail, M.; Dong, W.; Yang, Q. Machine learning driven smart electric power systems: Current trends and new perspectives. Appl. Energy 2020, 272, 115237. [Google Scholar] [CrossRef]
  35. Zheng, W.; He, Q.; Zhao, Y. Machine learning for battery research. J. Power Sources 2022, 549, 232125. [Google Scholar] [CrossRef]
  36. Pang, Y.; Li, Y.; Feng, Z.; Feng, Z.; Zhao, Z.; Chen, S.; Zhang, H. Forest fire occurrence prediction in China based on machine learning methods. Remote Sens. 2022, 14, 5546. [Google Scholar] [CrossRef]
  37. Mohajane, M.; Costache, R.; Karimi, F.; Pham, Q.B.; Essahlaoui, A.; Nguyen, H.; Laneve, G.; Oudija, F. Application of remote sensing and machine learning algorithms for forest fire mapping in a Mediterranean area. Ecol. Indic. 2021, 129, 107869. [Google Scholar] [CrossRef]
  38. Subramanian, S.; Crowley, M. Using Spatial Reinforcement Learning to Build Forest Wildfire Dynamics Models From Satellite Images. Frontiers 2018, 5, 6. [Google Scholar] [CrossRef]
  39. Tüfekci, P. Prediction of full load electrical power output of a base load operated combined cycle power plant using machine learning methods. Int. J. Electr. Power Energy Syst. 2014, 60, 126–140. [Google Scholar] [CrossRef]
  40. Kunxiang, D.; Zhang, Q.; Zhang, H.; Xiao, P.; Chen, J. Optimal Emergency Evacuation Route Planning Model Based on Fire Prediction Data. Mathematics 2022, 10, 3146. [Google Scholar] [CrossRef]
  41. Preeti, T.; Kanakaraddi, S.; Beelagi, A.; Malagi, S.; Sudi, A. Forest fire prediction using machine learning techniques. In Proceedings of the 2021 International Conference on Intelligent Technologies (CONIT), Hubli, India, 25–27 June 2021. [Google Scholar]
  42. Coffield, S.R.; Graff, C.A.; Chen, Y.; Smyth, P.; Foufoula-Georgiou, E.; Randerson, J.T. Machine learning to predict final fire size at the time of ignition. Int. J. Wildland Fire 2019, 28, 861–873. [Google Scholar] [CrossRef]
  43. Almuhaini, S.H.; Sultana, N. Forecasting Long-Term Electricity Consumption in Saudi Arabia Based on Statistical and Machine Learning Algorithms to Enhance Electric Power Supply Management. Energies 2023, 16, 2035. [Google Scholar] [CrossRef]
  44. Torp, S. Prediction of Battery Materials Properties with Machine Learning: Developing Algorithms to Discover Electrodes for Li-ion and Mg-Ion Batteries. Master’s Thesis, University of Oslo, Oslo, Norway, 2020. [Google Scholar]
  45. Flamenbaum, R.D.; Pompo, T.; Havenstein, C.; Thiemsuwan, J. Machine Learning in Support of Electric Distribution Asset Failure Prediction, (SMU Data Science Review 2019). Volume 2: No. 2, Article 16. Available online: https://scholar.smu.edu/datasciencereview/vol2/iss2/16 (accessed on 10 May 2025).
  46. Goodfellow, I.; Bengio, Y.; Courville, A.; Bengio, Y. Deep Learning; MIT Press: Cambridge, MA, USA, 2016; Volume 1, no. 2. [Google Scholar]
  47. Byoungjun, K.; Lee, J. A video-based fire detection using deep learning models. Appl. Sci. 2019, 9, 2862. [Google Scholar] [CrossRef]
  48. Ozcanli Asiye, K.; Yaprakdal, F.; Baysal, M. Deep learning methods and applications for electrical power systems: A comprehensive review. Int. J. Energy Res. 2020, 44, 7136–7157. [Google Scholar] [CrossRef]
  49. Sun, L.; Yan, J.; Qin, W. Path planning for multiple agents in an unknown environment using soft actor critic and curriculum learning. Comput. Animat. Virtual Worlds 2023, 34, e2113. [Google Scholar] [CrossRef]
  50. Viola, N. A Reinforcement Learning Approach to Real-Time Emergency Evacuation. 2024. Available online: https://www.diva-portal.org/smash/get/diva2:1891315/FULLTEXT01.pdf (accessed on 10 May 2025).
  51. Couto, G.C.K.; Antonelo, E.A. Generative adversarial imitation learning for end-to-end autonomous driving on urban environments. In Proceedings of the 2021 IEEE Symposium Series on Computational Intelligence (SSCI), Orlando, FL, USA, 5–7 December 2021. [Google Scholar]
  52. Codevilla, F.; Santana, E.; López, A.M.; Gaidon, A. Exploring the limitations of behavior cloning for autonomous driving. In Proceedings of the IEEE/CVF International Conference on Computer Vision, Seoul, Republic of Korea, 27 October–2 November 2019. [Google Scholar]
  53. Omran, S.; El Houby, E.M. Prediction of electrical power disturbances using machine learning techniques. J. Ambient. Intell. Humaniz. Comput. 2020, 11, 2987–3003. [Google Scholar] [CrossRef]
  54. Yu, N. Machine Learning and Big Data Analytics in Power Distribution Systems, (IEEE BDA Tutorial Series). IEEE Power & Energy Society. Available online: https://www.youtube.com/watch?v=4TdHoZYPk68 (accessed on 10 May 2025).
  55. Saad, M.H.; Said, A. Machine learning-based fault diagnosis for research nuclear reactor medium voltage power cables in fraction Fourier domain. Electr. Eng. 2023, 105, 25–42. [Google Scholar] [CrossRef]
  56. Dampage, U.; Bandaranayake, L.; Wanasinghe, R.; Kottahachchi, K.; Jayasanka, B. Forest fire detection system using wireless sensor networks and machine learning. Sci. Rep. 2022, 12, 46. [Google Scholar] [CrossRef]
  57. Sendek, A.; Cubuk, E.D.; Antoniuk, E.; Cheon, G.; Cui, Y.; Reed, E.J. Machine Learning-Assisted Discovery of Solid Li-Ion Conducting Materials. Chem. Mater. 2018, 31. [Google Scholar] [CrossRef]
  58. Sendek, A. Stanford ENERGY. Available online: https://www.youtube.com/watch?v=TSdgXlwIgtU (accessed on 10 May 2025).
Table 1. Virtual reality simulator models and areas of application.
Table 1. Virtual reality simulator models and areas of application.
Training Model AreaModels/Examples
Electricity and SafetyElectrical safety, VR training for electrical workers, power substation simulations, and electrical panel repairs [25,26,27,28,31,32]
Fire SafetyFire safety training in VR, modes of fire propagation and protection [24]
Student TrainingElectrical safety, VR training in electricity and magnetism, power electronics, and ElectroVR [22,23,29]
Specialist TrainingVirtual reality training system; technical skill improvement [30]
Table 2. Classification of ML algorithms used in different hazardous fields.
Table 2. Classification of ML algorithms used in different hazardous fields.
Algorithms for Work in a High-Risk Environment
Algorithms used for pathfinding and evacuation in a high-risk environmentDQN [33]SVM [34,35,36,37]
Q-Learning [34,38]
Neural Network/Artificial Neural Network [34,36,39,40,41]
Decision Tree [34,41,42]
Radial Basis Function [36,39]
SVR [35,43]
Random Forest [35,36,37,41,44,45]
Deep Learning [46,47,48]
Q network [33]
SAC [49]
PPO [50]
GAIL [51]
BC [52]
Algorithms used for fault preventionHMM [34]
Multi-Layer Perceptron [39]
Ant Colony Optimization [53]
BOA–NARX [43]
ARIMAX [43]
Linear Dimension Reduction [54]
Maximal Marginal Likelihood Estimation [54]
Multiclass SVM [55]
Algorithms used for fire forecast and preventionValue Iteration [38]
Policy Iteration [38]
Monte Carlo Tree Search [38]
Asynchronous Advantage Actor-Critic (A3C) [38]
Frequency Ratio-Multilayer Perceptron (FR-MLP) [37]
Linear Regression [56]
Algorithms used for finding new conductive elementsLogistic Function [57]
Supervised Learning [58]
Logistic Regression [58]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Vasilev, V.; Budakova, D.; Petrova-Dimitrova, V. Applications of Virtual Reality Simulations and Machine Learning Algorithms in High-Risk Environments. Eng. Proc. 2025, 100, 19. https://doi.org/10.3390/engproc2025100019

AMA Style

Vasilev V, Budakova D, Petrova-Dimitrova V. Applications of Virtual Reality Simulations and Machine Learning Algorithms in High-Risk Environments. Engineering Proceedings. 2025; 100(1):19. https://doi.org/10.3390/engproc2025100019

Chicago/Turabian Style

Vasilev, Velyo, Dilyana Budakova, and Veselka Petrova-Dimitrova. 2025. "Applications of Virtual Reality Simulations and Machine Learning Algorithms in High-Risk Environments" Engineering Proceedings 100, no. 1: 19. https://doi.org/10.3390/engproc2025100019

APA Style

Vasilev, V., Budakova, D., & Petrova-Dimitrova, V. (2025). Applications of Virtual Reality Simulations and Machine Learning Algorithms in High-Risk Environments. Engineering Proceedings, 100(1), 19. https://doi.org/10.3390/engproc2025100019

Article Metrics

Back to TopTop