Embedded Files
(Volume: 3, Issue: 7)
Speed Regulation In Induction Motors Mandated For Futuristic Applications
Induction motors find increased utilization in almost all industrial applications, as its cost-effective and simple design provides longer durability of operation with less maintenance. Owning the stator and the rotor as its key components, the operation of the induction motor begins with its stator generating a rotating magnetic field on the supply of power. This rotating magnetic field then induces magnetic flux and torque in the rotor, which causes it to rotate. The slip, meaning the speed of the stator’s rotating magnetic field minus the rotor speed, is the crucial factor needed to maintain the rotor in the rotating state. Though the operation might seem simple, there are also factors degrading the performance of the induction motors. The ability to not control the rotor speed at dynamic load conditions is one among the major drawbacks of the induction motor, seeking research attention. This can be explained by the following instances: A heavier load, needing increased rotor speed, reduces the slip that cause the rotor to speed down. Similarly, a lighter load, requiring moderate or slower rotor speed, increases the slip that cause the rotor to speed up. This simply implies that the induction motor operates at a fixed speed, making it ineffective for nonlinear and dynamic load applications. Hence, controlling the rotor speed, rotor flux or torque to support varying-load applications becomes mandatory. Generally, these motor parameters are optimally-controlled using Proportional-Integral-Derivative (PID) controllers or Sliding Mode Controllers (SMC). However, each of them provides distinctive advantages. While the PID controllers provide a linear control of the motor speed by calculating the error between the desired and the actual speeds, the SMC offers a non-linear control by letting the motor to slide to the desired state by switching between various control actions. Nevertheless, the steady state errors in the PID controller and the chattering phenomenon in SMC are a major concern in motor speed control. Hence, Arpit Yadav, Ranjay Das and Ganesh Roy from the Central Institute of Technology Kokrajhar, Kokrajhar, Assam, have investigated the hybridization of both PID and SMC controllers to regulate the induction motor speed, for use in both the linear as well as non-linear applications. As per their presentation in the 2024 IEEE International Students' Conference on Electrical, Electronics and Computer Science (SCEECS), the researchers have controlled the torque and the rotor flux using SMC, while employing the PID controller for motor speed regulation. The researchers, though finding their investigation results as outperforming, instigate the need for chattering removal using higher order SMCs. Since the induction motors can support varied applications in the transportation, energy, robotics, housing, healthcare and similar other sectors, motor speed regulation requires extensive study in the near future. Image courtesy: www.vecteezy.com
Optimized Onboard Machine Learning Models Mandated For Real-Time Applications
Machine learning with deep learning approaches is revolutionizing almost all applications, owing to its ability to learn from raw data with improved generalization and scalability. However, this approach might also lose its efficacy in real-time applications, which usually involve resource-constrained devices requiring incessant learning. It is the additional time and computational complexity, which is associated with the re-training of the deep model from scratch to learn new classes in real-time, that causes this inefficacy. TinyML (Machine Learning) with Class Incremental Learning (CIL) is a novel machine learning paradigm, which enables the resource-constrained devices, like microcontrollers, sensors or other embedded systems, to learn new classes onboard directly, without depending on a central or cloud server to meet the memory constraints. Though the approach provides a secure, real-time learning and decision making with the locally-available data on the devices itself, catastrophic forgetting is a critical issue, needing research attention. Catastrophic forgetting means that a machine learning model fails to retain the past knowledge, when new classes are introduced for training. Since the real-time applications involve highly non-stationary data with devices involving limited resources, catastrophic forgetting questions the reliability of the model itself. Replay-based method is one among the methods, which could alleviate catastrophic forgetting to a certain extent. By this method, a small portion of the past knowledge is buffered and whenever a new class is encountered, the model will be re-trained with both the buffered as well as the new data. However, this knowledge buffering also impacts the resources available onboard. Hence, optimizing the TinyML models to account for catastrophic forgetting and power or other resource constraints looks promising. Suraj Kumar Pandey and Shivashankar B. Nair, two researchers from IIT, Guwahati, Assam, India, have optimized the TinyML models using Genetic Algorithm (GA) to facilitate onboard CIL in microcontrollers, so as to perform gesture recognition with time-series data. In their article in the GECCO '24 Companion: Proceedings of the Genetic and Evolutionary Computation Conference Companion, the researchers have proven the competency of their optimized model to operate in resource-constrained settings. Since the impending technologies, for instance, automated vehicles, industrial IoT equipment, robotic machineries and wearable technology-based devices all operate with limited resources, a deeper investigation on onboard TinyML or similar other models and their optimization is highly mandated in future. Image courtesy: www.freepik.com
Sign Language Interpreters For The Hearing-Impaired
Several languages exist all around the globe and at times, conversing each of them in the same region too varies. So, what can a visitor with no knowledge on the native language do? Speak verbally through gestures? Unfortunately, verbal communication too changes as does the language in a region change, making people to struggle to establish proper communication. If it is the case, then it will be definitely onerous for a person with hearing impairment, who naturally convey or understand the spoken language using gestures. It is at this instant, where the sign language interpreters seem to render hands. Sign language interpreters are devices, which could achieve textual translation between the sign language and the spoken language or conversely, as per the user’s need. It is the wearable technology and the gesture as well as the speech recognition systems, which have significantly influenced the advancements in sign language interpreters. P Kumar, S Senthil Pandi, L Priya and V Rahul Chiranjeevi from Rajalakshmi Engineering College, Chennai, India have developed a mobile app that employed the Inception v3 architecture to perform the gesture classification of a user, followed by its translation into textual information. The researchers have chosen the Inception v3 architecture, owing to its remarkable performance in the ImageNet Large Scale Visual Recognition Challenge (ILSVRC). Presenting their sign language interpretation model in the 2024 International Conference on Communication, Computing and Internet of Things (IC3IoT), the researchers have effectively provided a simple, hand-held, mobile device for sign language interpretation to the people with hearing impairments. However, the accurate translation of gesture into textual information mandates further research, as the tone, the context and the meaning cannot be interpreted rightfully always, especially in real-time applications. Being an embodiment of Human-Computer Interaction (HCI) and its wide usage in education, healthcare management, tourism and travel, media and entertainment, and similar other applications, research on sign language interpretation has a great scope in this digitally-transforming era. Image courtesy: www.vecteezy.com
Raise Online-Business Revenue By Analyzing Shopper’s Intention With Data Mining Approaches
Consumer nowadays show extreme interest in shopping online, owing to the convenience that it provides in shopping anything, anywhere and at any time, along with the ease of product delivery at doorstep. In fact, online shopping allows the consumer to access a wider range of products and to make comparisons on the prices or the discounts on a product from different retailers. However, this convenience created to the consumers can possibly hinder the revenue generated by the retailers, especially when they sell the same product. Hence, understanding the shopper’s intention to buy a product online has a crucial role in increasing the retailers’ revenue or to optimally-plan their business strategies. But, how can the users’ intention on product buying be predicted? It is through analyzing the clickstream data, which represents the information being tracked from the user browsing the website like, the web pages or the web links visited, the time spent on it etc. Further, it is also obvious that this prediction involving voluminous data cannot be manually-handled that the data mining or the machine learning approaches could only effectively-aid at this juncture. Sakshi Katara, Chandresh Kumar Karn, Manvi Agrawal, Devendra Jamaliya, Ishika Mittal and Dr. Ankush Verma have attempted to classify the shopper’s intention in purchasing the product and to predict the revenue, using the session as well as the visitors’ information being tracked, as the customers surfed the web. The authors have compared various supervised machine learning algorithms and ensemble data mining methods for this purpose in their article in the International Journal of Latest Trends in Engineering and Technology, vol.19 (2), mentioning Random Forest to outperform the other algorithms in accurately predicting the revenue. However, there are also chances to improve the prediction accuracy in the real-time scenario with novel algorithms and considering additional clickstream data. As the craze on online shopping is never going to have an end, developing business prediction systems to improve the customer’s shopping experiences or to plan the marketing strategies for increasing the retailers’ revenue will certainly have a great impact on the e-commerce industry. Image courtesy: www.vecteezy.com
EMI Shielding Using Porous Carbon Composites With Iron Concentrations
Carbon composites are greatly prioritized in applications requiring high electromagnetic shielding, such as military, automotive, aerospace, healthcare, consumer electronics and many more. Despite its higher conductivity and its ability to reflect, absorb or dissipate Electromagnetic Interferences (EMI), the huge cost associated with the manufacturing of carbon fibers has necessitated the carbon manufacture from organic and sustainable materials, like the coconut coir. The Coconut Coir-based Porous Carbon (CCPC) is a highly-porous carbon composite, which is obtained by letting the fibrous material on the outer coconut husk (known as the coir) to undergo the pyrolysis process. Though CCPC involves porous carbon manufacture with reduced costs, its electromagnetic shielding effectiveness does not equal the shielding effectiveness of carbon fiber composites or other metal composites. Hence, K Robin Johny, C Bhagyanathan and J David Rathnaraj from Sri Ramakrishna Engineering College, Coimbatore, India, have investigated the use of iron (Fe) particles in four varying proportions to improve the electromagnetic shielding effectiveness of the CCPC composites. Making a comprehensive analysis on the Fe-infused CCPC composite in their article in the Proceedings of the Institution of Mechanical Engineers, Part L: Journal of Materials: Design and Applications, Sage, the researchers have found that the EMI absorption was more distinct with increasing Fe concentrations. Though the researchers assured a lower reflection loss with 99% electromagnetic wave attenuation with their composite material, a smaller structural degradation still occurred with increasing iron concentrations. Apart from this research, it is a known fact that iron is a heavy-weight, corrosive metal, which can even interfere with high-frequency electromagnetic waves. Hence the research on cost-effective and sustainable alternative composites, yielding improved EMI shielding, are still sought in the future.
Image courtesy: www.vecteezy.com
Page updated
Google Sites
Report abuse