In the 21st century, researchers have been showing keen interest in the areas of wireless networking and internet of things (IoT) devices. Conventionally, batteries have been used to power these networks; however, due to the limited lifespan of batteries and with the recent advancements in piezoelectric technology, there is a dramatic increase in renewable energy harvesting devices. In this research, an eco-friendly wind energy harvesting device based on the piezoelectric technique is analytically modeled, numerically simulated, and statistically optimized for low power applications. MATLAB toolbox SIMSCAPE is utilized to simulate the proposed wind energy harvester in which a windmill is used to produce rotational motion due to the kinetic energy of wind. The windmill’s rotational shaft is further connected to the rotary to linear converter (RLC) and vibration enhancement mechanism (VEM) for the generation of translational mechanical vibration. Consequently, due to these alternative linear vibrations, the piezoelectric stack produces sufficient electrical output. The output response of the energy harvester is analyzed for the various conditions of piezoelectric thickness, wind speed, rotor angular velocity, and VEM stiffness. It is observed that the electrical power of the proposed harvester is proportional to the cube of wind speed and is inversely proportional to the number of rotor blades. Furthermore, an optimization strategy based on the full factorial design of the experiment is developed and implemented on MINITAB 18.0 for evaluating the statistical performance of the proposed harvester. It is noticed that a design with 3 rotor-blades, having 3 mm piezoelectric thickness, and 40 Nm⁻¹ stiffness generates the optimum electrical response of the harvester.

Vibration attenuation is necessary in many electro-mechanical systems. In this paper electromagnetic damping is applied on a flexible cantilever beam (FCB) made of ferromagnetic material. The lateral vibration is controlled by employing two electromagnetic poles mounted in the same axis as that of a vibrating cantilever beam. A power amplifier is developed to provide desired current to the electromagnet. Comparison of damping achieved at different electromagnetic forces is presented. The study employs a new method of applying a constant DC power to one of the electromagnets while other is connected to a variable current source. Experimental results are presented that show the effectiveness of the presented approach.

Blade faults are among the most prevalent and potentially catastrophic failures in gas turbines and compressors. Early detection of these faults is critical to ensure efficient and safe operation. Unlike bearing and gear faults, which exhibit standard fault characteristics that can be readily analyzed through time and frequency domain methods, diagnosing blade faults presents unique challenges. The complex structure of rotor systems and the presence of noise obscure the discriminative features of blade faults in vibration signals, necessitating the expertise of highly experienced specialists for accurate interpretation. This paper presents a novel hybrid deep learning model based on convolutional neural network (CNN) and transformers to address these challenges in diagnosing the most frequent blade faults such as blade tip rubbing, blade stage rubbing, twisted blade, and loss of blade part. The study conducted on a multistage rotor system designed and developed within our laboratory demonstrates a significant advancement in blade fault diagnosis. The developed model achieved a diagnostic accuracy of over 90.45%, validated by confidence interval analysis even with a challenging noisy condition for various stages and fault combinations. The model demonstrates its robustness by accurately identifying blade faults for various fault combinations, surpassing baseline CNN (52.36%–79.87%) and transformer (58.60%–74.27%) models with minimal human intervention and signal processing requirements. The hybrid CNN–transformer′s low parameter footprint contributes to balanced computational needs, enabling its practical application in real‐time scenarios. This powerful hybrid algorithm integrates the strengths of both CNN and transformer architectures, vital for intricate blade fault diagnosis. The CNN initially processes raw vibration sensor data, extracting local, translation‐invariant features like specific vibration patterns indicating localized blade damage. This acts as an automated feature engineering front‐end. These features then feed into the transformer, which uses its self‐attention mechanism to understand long‐range dependencies and global context. For blade fault diagnosis, this enables it to connect localized features to the overall blade health, capture complex temporal relationships for evolving faults, and prioritize important signal segments. This innovative combination significantly enhances the reliability and efficiency of blade fault diagnosis in complex rotor systems.