Publications
Published on August 2025
Vibration Institute Annual Training Conference recently featured Drumil Joshi, Monitoring & Diagnostics Analyst at Southern Power, along with co-authors Christopher Harrison and Andrew Riley, for presenting VIBRIS: Vibration Intelligence Bearing Reliability Integrated System—a next-generation platform for wind turbine condition monitoring:contentReference[oaicite:0]{index=0}.
VIBRIS introduces a multi-layered, secure data acquisition framework with token-based authentication, hierarchical retrieval, and proprietary HEX encoding to safeguard turbine sensor data. Integrated with PI Server and SCADA streams, the system applies ensemble anomaly detection—using Isolation Forest, Local Outlier Factor, One-Class SVM, and K-Means Clustering—to separate transient disturbances from persistent mechanical deviations:contentReference[oaicite:1]{index=1}.
“VIBRIS transforms how we approach turbine health—moving from reactive maintenance to predictive intelligence,” Joshi emphasized. “By securing data at every step and layering advanced machine learning, we can detect faults long before they escalate.”
Conference results highlighted VIBRIS’s ability to forecast and pre-detect maintenance issues, setting a new benchmark for reliability assessment. Its ensemble framework demonstrated superior accuracy in identifying gearbox imbalances, bearing misalignments, and long-term degradation trends across wind turbines:contentReference[oaicite:2]{index=2}.
As recognized by the Vibration Institute, VIBRIS represents a paradigm shift in predictive maintenance—where advanced security, vibration analytics, and AI-driven automation converge to deliver smarter, safer, and more resilient renewable energy operations.
Published on June 2025
IEEE PVSC (Photovoltaic Specialists Conference) recently featured Drumil Joshi, Monitoring & Diagnostics Analyst at Southern Power, for co-authoring a groundbreaking presentation titled “Using Open-Source Forecasts for Solar Plant Maintenance Outage Scheduling Can Reduce Lost Energy.”
The study demonstrates how integrating open-source solar forecasts into outage planning allows operators to strategically schedule maintenance during low-generation periods. By aligning downtime with predicted irradiance patterns, the approach minimizes lost energy and improves overall solar plant efficiency.
“Forecast-driven scheduling enables operators to be proactive—using free and accessible data to cut losses, optimize performance, and support cleaner energy delivery,” Joshi explained.
The work emphasizes practical innovation, providing not only methodology but also open reference code, ensuring other utilities can replicate and scale the solution across renewable fleets.
Authored in CXOToday – 21st March 2025
Drumil Joshi, a globally recognized expert in AI, automation, and creative intelligence, is reshaping the narrative around artificial intelligence. In his recent feature with CXOToday, Joshi explores the AI Paradox—how automation, instead of replacing human creativity, is fueling a renaissance of innovation, imagination, and purpose-driven work.
“AI isn’t eliminating creativity—it’s amplifying it,” said Joshi. “By automating repetitive tasks, we’re freeing up the most human part of our minds—the ability to think, ideate, and innovate.”
Joshi’s forward-thinking perspective challenges traditional fears around AI. His work demonstrates how intelligent automation can empower individuals and teams across sectors to focus on strategic, imaginative, and impactful pursuits. Through real-world case studies and research, he shows how AI systems are complementing rather than competing with human intelligence.
- Accelerated innovation cycles in R&D and product development
- Enhanced human-AI collaboration for strategic problem solving
- Creative liberation for designers, developers, and analysts
“The synergy between human insight and machine precision is the new creative frontier,” Joshi explained. “This era isn’t about humans versus AI—it’s about co-evolution. When we align AI tools with our curiosity and vision, we unleash a new age of creative possibility.”
Authored in Hindustan Times – 10th Feb 2025
Drumil Joshi, a leading expert in AI and data science for renewable energy, is driving the transformation of the clean energy sector. In his latest feature with Hindustan Times, Joshi explores how AI and data analytics are reshaping renewable energy operations, making them more efficient, predictive, and resilient.
“Data science and AI are no longer optional in the energy sector—they are essential for optimizing performance and ensuring a stable, sustainable grid,” said Joshi. “By leveraging machine learning, we can enhance predictive maintenance, improve asset reliability, and maximize renewable energy output.”
His work focuses on advanced AI-driven analytics for wind, solar, and battery energy storage systems (BESS), leading to significant industry advancements:
- Increased efficiency in wind and solar asset performance
- Real-time anomaly detection for predictive maintenance
- Optimized grid stability through AI-powered forecasting models
“The renewable energy revolution is powered by AI. The ability to process vast amounts of data in real time enables energy operators to make proactive, data-driven decisions,” Joshi explained. “This is not just about efficiency; it’s about ensuring a resilient and sustainable energy future.”
Published on January 12, 2025
TechBullion recently featured an insightful article authored by Drumil Joshi, a renowned AI and renewable energy expert, titled “Revolutionizing Renewable Energy: Drumil Joshi on Predictive Grid Harmony, AI, and the Future of Grid Stability.”
In this piece, Joshi outlines how AI-powered forecasting can transform grid operations by enabling smarter, data-driven decisions. He explains how predictive analytics can help balance the volatility of solar and wind power, improving overall grid efficiency and stability.
“We’re moving from reactive to proactive energy systems,” Joshi writes. “AI allows grid operators to forecast fluctuations, reduce instability, and enhance real-time decision-making.”
Joshi envisions a future where artificial intelligence and real-time data work hand-in-hand to create a resilient and sustainable energy infrastructure. His article calls for widespread adoption of predictive models to ensure harmony between renewable energy inputs and grid performance.
The feature is part of Joshi’s broader mission to integrate AI into the heart of the renewable energy transition, ensuring cleaner, smarter, and more stable power systems for the future.
Published on December 14, 2024
Drumil Joshi, a trailblazer in AI and renewable energy, has co-authored a groundbreaking research paper presented at the 2024 IEEE ICERCS Conference, introducing a next-gen AI model for solar energy harvesting in IoT systems.
The study, titled “Intelligent Solar Energy Harvesting and Management in IoT Nodes Using Deep Self-Organizing Maps,” showcases a powerful solution for making smart devices energy self-sufficient—achieving an outstanding 91.3% prediction accuracy and the fastest execution time (5624 ms) among all compared models.
“We’re not just forecasting energy—we’re creating devices that can think, adapt, and manage power on their own,” Joshi noted.
Powered by the Extended Deep Self-Organizing Map (DSOM) algorithm, the model outperforms SVM, SOM, and GRU, enabling real-time solar prediction for sectors like smart cities, agriculture, and environmental monitoring.
Visuals in the study (see Figure 5 & 6, page 5) reinforce the model’s superior performance across accuracy and efficiency metrics.
Now leading energy analytics at Southern Power, Joshi continues to drive innovation at the intersection of AI, sustainability, and intelligent energy systems, redefining how the world powers its future.
Drumil Joshi Unveils Breakthrough Vision in AI-Powered Acoustic Diagnostics for Solar and BESS Systems
In his visionary article titled “Tuning the Silent Symphony,” AI and renewable energy expert Drumil Joshi explores the transformative potential of AI-powered acoustic diagnostics in the real-time monitoring and health assessment of solar photovoltaic (PV) systems and battery energy storage systems (BESS).
Joshi highlights a critical gap in traditional inspection methods, which often detect faults only after substantial damage has occurred. In contrast, acoustic diagnostics “listen” to the subtle sound signatures emitted by energy systems, identifying issues like microcracks and internal material stresses at an early stage—long before physical symptoms emerge.
He further details the integration of ultrasonic sensing with lightweight neural networks, enabling swift and accurate diagnosis across vast renewable infrastructure. Joshi envisions a near-future powered by drone swarms, digital twins, and AI-driven simulation tools that will elevate asset reliability, safety, and performance in ways previously thought impossible.
In his article titled “Mood as a Metric,” Drumil Joshi presents an innovative approach to renewable energy diagnostics through the development of an AI-powered Mood Ring Dashboard. This futuristic system transforms complex operational data from solar panels, wind turbines, and battery storage systems into intuitive, color-coded indicators—green for optimal performance, amber for caution, and red for critical issues—enabling operators to respond swiftly and effectively.
The dashboard is powered by a machine learning engine trained on extensive operational datasets, including turbine vibrations, inverter waveforms, panel temperatures, battery charge profiles, and weather anomalies. It computes a Real-Time Mood Index (RTMI), assigning each asset a dynamic “mood” that visually communicates system health and emerging anomalies.
Field trials in the U.S. and India showed measurable improvements:
+19% increase in fault detection speed
+23% faster decision-making by junior operators
-11% reduction in reactive maintenance costs within just four months
These results underscore the dashboard’s potential to enhance operational efficiency and reduce downtime in renewable systems.
In his article titled “The Rise of the Renewable Swarm,” Drumil Joshi introduces a groundbreaking approach to renewable energy management by applying swarm intelligence principles—drawn from the collective behaviors of bees, ants, and birds—to wind turbines and solar panels.
Joshi envisions a future where renewable energy assets operate as adaptive, cooperative systems, responding in real time to environmental conditions and grid demands. This decentralized model contrasts with traditional centralized control systems, offering enhanced efficiency, resilience, and autonomy.
At the core of this approach is a four-layer framework:
Smart Assets: Each turbine or panel functions as an intelligent node, capable of edge-based decision-making using lightweight AI, monitoring local conditions like temperature, wind speed, irradiance, and mechanical stress.
Communication Mesh: Assets communicate through peer-to-peer protocols such as MQTT or LoRaWAN, sharing insights with neighboring units without relying on a central hub.
Swarm Coordination Algorithms: Inspired by nature and refined through AI research, these algorithms enable assets to collectively adapt and make decisions without conflict.
Learning and Memory: Over time, assets evolve by learning from past behaviors, storing event-response patterns similar to how ants reinforce trails.
This innovative system allows for scenarios where, for example, wind turbines adjust torque based on neighboring units’ experiences, or solar panels collectively tilt to prevent overgeneration, all without human intervention.
Joshi’s insights are supported by real-world applications: a 2022 study in IEEE Transactions on Smart Grid demonstrated that agent-based coordination in distributed solar systems led to an 18% boost in operational efficiency during high-variability hours.
Drumil Joshi Co-Authors Breakthrough Research on 3D AR Visualization for Smart IoT Systems
Drumil Joshi, alongside researchers from Dwarkadas J. Sanghvi College of Engineering, has co-authored a visionary research paper exploring Augmented Reality (AR) and Computer Vision to enable real-time 3D visualization in IoT environments.
The paper, titled “Delve into the Realms with 3D Forms,” introduces a compact, cost-effective system using Raspberry Pi and OpenCV to detect objects, match features, and augment 3D images onto real-world scenes in real time.
“Our goal was to create a smart, scalable platform for industries like retail, gaming, medicine, and electronics using lightweight hardware and robust algorithms,” said Joshi.
The system leverages: ORB (Oriented FAST and Rotated BRIEF) for high-speed feature detection. FLANN (Fast Library for Approximate Nearest Neighbors) for accurate feature matching. 3D projection algorithms for augmented visualization directly on live video feeds
The solution is fully Python-based, highly adaptable, and ideal for edge-based AR applications, extending use cases in smart cities, industrial design, healthcare, and intelligent retail.
“We’ve proven that real-time 3D AR can be achieved using low-cost hardware without sacrificing precision,” Joshi added.
This work stands as a practical blueprint for bringing immersive AR experiences into everyday IoT systems, bridging the digital and physical worlds in a seamless, scalable way.
Published in June 2021 | Turkish Online Journal of Qualitative Inquiry (TOJQI), Volume 12, Issue 6
Drumil Joshi, in collaboration with a multi-institutional team, has co-authored a research paper introducing a robust AI-based model for predicting customer churn in the telecom sector. Titled “AI Based Nose for Trace of Churn in Assessment of Captive Customers,” the paper offers a machine learning framework capable of identifying potential customer drop-offs with remarkable precision.
The study tested eight ML algorithms, including Random Forest, Gradient Boosting, and XGBoost, with extensive hyperparameter tuning to optimize performance. The XGBoost Hyper-Parameterized Classifier emerged as the best model, achieving an accuracy of 81.13% and precision of 80%, while maintaining low false positive rates.
“Understanding and predicting churn is critical for sustainable customer retention. Our AI framework not only detects churn early but also explains the decision with interpretable metrics,” Joshi noted.
The research used real-world telecom data, applying advanced boosting techniques, visual analytics (as shown in Figures 5 & 6), and evaluation metrics such as ROC, Gini coefficient, and odds ratio. The final model was also assessed for real-world viability using confusion matrices and feature importance analysis.
Joshi’s work presents a scalable and actionable approach to customer retention strategy—delivering value to both businesses and data scientists in churn-heavy industries.
Published on April 5, 2021 | Turkish Journal of Computer and Mathematics Education, Vol. 12, No. 3
Drumil Joshi, alongside researchers from Mumbai University, has co-developed a groundbreaking AI-powered solution for smart water management, aimed at reducing water wastage and optimizing consumption through intelligent forecasting and automation.
Titled “Innovative Smart Water Management System Using Artificial Intelligence,” the research proposes a low-cost, plug-and-play IoT-based system that uses machine learning and time series analysis to predict household water usage. The system tracks daily, weekly, and monthly water consumption via sensors and cloud-connected ESP32 microcontrollers, enabling real-time monitoring through a mobile app built on MIT App Inventor.
“Our goal is to make water conservation smarter, more accessible, and predictive,” said Joshi. “The system can alert households of leaks, reduce overuse, and even forecast usage for the next 5 days using SARIMAX time series models.”
Key ML techniques like Gradient Boosting, Random Forest, and Lasso Regression were deployed to analyze consumption trends. With detailed hyperparameter tuning and visual dashboards, the system achieved forecast accuracy above 85% and supports applications in residential, industrial, and municipal settings.
The paper also outlines future applications in food processing and chemical plants, where smart fluid dispensing can enhance efficiency and safety.
This research marks a significant step toward data-driven sustainability, where AI and IoT converge to protect one of Earth’s most critical resources—water.
An Efficient Supervised Machine Learning Model Approach for Forecasting of Renewable Energy to Tackle Climate Change
This paper presents a machine learning model to accurately forecast renewable energy usage in the EU, focusing on solar and wind energy. The approach uses live data from the ENTSOE transparency platform, utilizing algorithms such as linear regression, random forest, and support vector machines. The system is deployed via a user-friendly web app, allowing users to select a country and time span to predict energy consumption trends. The model’s precision is validated by achieving low SMAPE values (1-2%), making it reliable for real-time decision-making and climate strategy. This tool aims to support the shift towards carbon neutrality by 2050, aligning energy forecasting with climate goals.
Elements of Nature Optimized into Smart Energy Grids using Machine Learning
This paper explores the use of machine learning to predict solar energy output by analyzing data like temperature, humidity, and wind direction. Using data from a weather station in Hawaii, the study applies algorithms such as linear regression, random forest, and support vector machines to predict solar radiation patterns, helping to balance energy supply in smart grids. The best-performing model was a hyper-parameterized random forest with a cross-validation score of 72%, attributed to its ability to accurately track daily cycles. The research highlights how machine learning enhances solar energy integration by making predictions more reliable, aiding the shift toward sustainable energy systems.