Diffusion-driven adaptive radiance refinement with PPO-based optimization for robust solar irradiance forecasting Natrayan L, Chenga Reddy Peddamangari, M Prem Kumar Reddy, Seeniappan Kaliappan, Ramya Maranan, Anand Rajendran Results in Engineering, 2026 • Introduces a novel fusion of diffusion-based refinement and reinforcement learning to enhance solar irradiance forecasting and PV control. • PPO-based controller continuously self-corrects prediction errors in real time, ensuring adaptive and resilient PV system performance. • Achieves around 8% improvement in prediction accuracy and notable reduction in forecast volatility compared to advanced deep learning and hybrid models. • Delivers a scalable, real-time, and autonomous forecasting-and-control framework suitable for smart grid and distributed solar energy applications. Rapid global transition toward renewable energy sources has amplified the importance of accurate solar irradiance forecasting for reliable PV system operation. Conventional models such as LSTM, ANN, and optimization-based hybrids face challenges in capturing nonlinear irradiance variations, mitigating atmospheric noise, and adapting to sudden weather changes. To address these limitations, this study introduces a novel Diffusion-Based Adaptive Radiance Refinement Layer (ARRL) integrated with Proximal Policy Optimization (PPO) to enhance denoising, adaptivity, and real-time error correction. The ARRL suppresses stochastic irradiance noise, while PPO continuously refines prediction outputs through reinforcement-guided optimization. The proposed framework was implemented in Python using TensorFlow 2.15 and evaluated on the Kaggle Solar Irradiance and Weather Forecasting Dataset, consisting of 1000 temporal records sampled at 30-minute intervals. Experimental results reveal that the proposed Diffusion and PPO model achieved a MAE of 10.28 W/m², RMSE of 14.17 W/m², MAPE of 2.25 and an R² score of 0.97, representing an approximate 8% improvement in prediction accuracy compared to advanced LSTM and hybrid optimization benchmarks. Moreover, the framework demonstrated robust generalization under cloud-induced fluctuations and significant reductions in forecast volatility. The synergistic fusion of diffusion denoising and reinforcement-based adaptive learning delivers a self-correcting and scalable solution for solar forecasting. In conclusion, the proposed model establishes a highly interpretable and resilient architecture, setting a new direction for intelligent, autonomous, and data-driven solar irradiance prediction systems in dynamic environmental conditions.
Ultrasound supported synthesis of waste mangifera indica linn biodiesel: an optimization using whale algorithm S Arumugam, Chengareddy Peddamangari Venkatesulu Reddy, Tamilarasan Arulvalavan, Sriram Gopalasamy Energy Sources Part A Recovery Utilization and Environmental Effects, 2025 The synthesis of biodiesel from vegetable oil is not the right choice as its feedstock and production cost is expensive than diesel. To address this, the study reports the implementation of Whale Algorithm (WA) for the optimization of biodiesel synthesis from cost-effective waste feedstock, i.e. waste mango seed kernel oil (WMS) using ultrasound-supported methanolysis process. The influence of methanolysis process variables, i.e. methanol:WMS molar ratio (12:1–20:1), sonication time (10–30 min), heterogeneous catalyst concentration (1–3 wt.%), and ultrasound amplitude (25–75%) on the yield percentage of WMS methyl ester were examined and optimized. The WA model is used to augment the yield percentage upto 90.88% at optimum methanol:WMS molar ratio of 16.32:1, the sonication time of 19.75 min, the quantity of heterogeneous catalyst of 2.28 wt.% and an ultrasound amplitude of 54.31%. 1H-NMR spectral approach was adopted to ensure the formation of WMS methyl ester. The combined use of heterogeneous catalyst along with ultrasonication significantly improved the conversion percentage of WMS methyl ester to 90.88% over the conventional transesterification process yield of 83%. The outcome of this study confirmed that the optimization by WA showed marginal improvement in yield percentage of WMS methyl ester as compared to those obtained using response surface methodology (RSM), a commonly applied optimization method.
Machine Learning Techniques Applied in Predictive Maintenance: A Review P. Chenga Reddy, Karamala Naveen, Naveen Kilari, Nagendra Panini Challa AI and Machine Learning for Mechanical and Electrical Engineering, 2025 The people and the society were fully dependent on industries since the first mankind to work on it from the first Industrial Revolution to the fourth Industrial Revolution i.e. Industry 4.0. In every industry revolution, tremendous efforts were taken by the researchers and inventors to incorporate and implement the technology for the performance improvement. In otherward, say that Industry 4.0 is directly cooperating with the scientific revolution. In Industry 3.0 revolution itself, many technologies were introduced like electronics, information technology, and automation. But it is the time to introduce that machine learning (ML) and artificial intelligence (AI) for the betterment of the performance and maintenance. Decision-making for the maintenance with machines is the biggest challenge in industry which involves enormous input of data and customization in the engineering process. It has been noted that Industry 4.0 surveys and tutorials mostly cover data analytics and ML techniques to modify manufacturing processes, leaving out predictive maintenance (PdM) techniques and their implementation. Although it hasn’t been thoroughly studied, ML technologies have a lot of potential in the broad field of mechanical engineering to reduce rejection, enhance product superiority, system optimization, etc. The main commitment of this work is to provide an organized assessment of works that use vision systems and to entice scholars to examine its possible integration with ML techniques from the perspective of Industry 4.0. At the appropriate points, current developments and research gaps in the field of manufacturing in conjunction with ML are also covered.
Heart illness forecast using a hybrid transfer machine learning classifier integrated with a deep feature extractor Nirmala M, Bhargavi Krishna, Chengareddy P, Naveen Karamala 3rd International Conference on Emerging Computation and Information Technologies Icecit 2025 Book of Abstracts, 2025 CVDs (cardiovascular diseases) continue to be a major global cause of death, primarily due to the late detection of cardiac abnormalities. Advance with accurate forecast of heart illness is therefore essential for preventing cardiac arrest and enhance patient outcomes. Although machine learning (ML) methods is revealed ability in automated analysis, most existing approaches depend on single-source data and rely on isolated techniques, limiting their predictive capability. In order to improve the precision and reliability of heart attack forecasts, this investigation introduces a novel hybrid diagnostic framework that combines several machine learning models with an advanced feature generator. Statistical representations are extracted from the dataset using deep learning models (DL_Simple and DL_Complex). These representations are then assessed using a number of machine learning methods, such as Random Forest, XGBoost, Support Vector Machine (SVM), Logistic Regression (LR), and Gradient Boosting. To choose the best classifier, a thorough performance evaluation is carried out. Results from experiments show that the suggested multidisciplinary architecture greatly increases prediction accuracy. Logistic regression outperformed other cutting-edge methods among the assessed models, showing an overall precision of 92.11% and an extent beneath the ROC curve (AUC) of 96.59%.
A review on synthesis, combustion, performance, and emissions of compression ignition engine fueled by waste cooking oil-based biodiesel Chengareddy P, Naveen Kilari, Arumugam S Egyptian Journal of Petroleum, 2024 This decade has seen a significant concentration on the development of sustainable and renewable products to replace fossil fuels in industry, science, and the environment. Another problem is that waste cooking oil (WCO) disposal damages the environment. Using WCO to produce biofuels is the answer to both of these problems. This paper reviews the ideal method and conditions for the synthesis of biodiesel from WCO via transesterification. Studies on the pretreatment WCO are also reviewed, as biodiesel is gradually becoming a more viable alternative to sustainable and renewable fuel. The impact of biodiesel on engine performance and emissions is highlighted in this review. The results of performance tests conducted on Compression Ignition (CI) engines using various blends of biodiesel confirm that certain blends are as efficient as diesel while also reducing emissions. This review also looks at how biodiesel and nanoparticles affect engine performance and emissions, with favorable findings. A proper way to address waste management and energy generation is to use WCO to make biodiesel.
Vegetable Oil-Based Pentaerythritol Ester for Industrial Air Compressor Applications: A Tribological Investigation P. Chengareddy, S. Arumugam Tribology in Industry, 2023 Biolubricants are being developed more frequently as a result of environmental issues, which also improve performance and lessen friction and wear. This study seeks to examine the effects of commercial compressor oil-SAE30, and its blend with pentaerythritol ester (PE) derived from vegetable oil on tribological properties. Two sequential transesterification processes were used to create the pentaerythritol ester made from vegetable oil utilizing the ultrasonic irradiation method. Using an air compressor lubricant, load, and sliding speed for each experimental run, a pin-on-disc tribometer is utilized to examine the friction and wear characteristics of the material. According to the experimental findings, PE 75 (75 vol.% - PE and 25 vol.% - SAE30) performs better than PE and SAE30 in terms of minimum coefficient of friction (COF) and specific wear rate (SWR) under high loads and high sliding speeds. Furthermore, scanning electron microscope (SEM) and Energy Dispersive Analysis of X-Ray (EDAX) analyses were used to examine the morphology behavior of the pin surfaces. Due to the coating that covers the specimen's surface, PE 75 compressor oil displays smoother surfaces than other lubricated surfaces.
Vegetable Oil Based Compressor Oil-optimising of Tribological Characteristics P. Chengareddy, Arumugam Shanmugasundaram Australian Journal of Mechanical Engineering, 2023 This study explored the use of vegetable oil-based polyolester of rapeseed oil as a potential biolubricant for reciprocating air compressor. Rapeseed oil-based pentaerythritol ester (PE) was formulated via ultrasound-assisted transesterification process. Response surface methodology (RSM) based D-optimal design of experiments was used to analyse the tribological parameters, such as disc rotational speed, load and compressor oils on various tribological behaviours, namely, coefficient of friction (COF) and specific wear rate (SWR) using a pin-on-disc tribometer. The least specific wear rate of 1.85 × 10−5 mm3/Nmand COF of 0.0393 was recorded at optimal conditions. The surface morphology analysis of the tested pin using SEM/EDX and tribo-film formation analysis of disc surface using XPS technique disclosed that blended compressor oil (PE75), i.e., 75% by volume of PE and 25% by volume of synthetic compressor oil (SAE30) has numerous capabilities as environmental friendly compressor oil.
