Proceedings of the 7th International Conference on Communications and Cyber Physical Engineering

Table of Contents

1. Fault Detection in Wind Turbine Using IoT

   K. V. Dhanalakshmi, G. Naga Mallika, E. Sai Sruthi, V. Prathika, R. Susmitha, G. Bhargavi

   Abstract

   With the rapid expansion of wind energy generation, ensuring the efficient operation and maintenance of wind turbines has become paramount. This paper puts forward a comprehensive approach for fault detection in wind turbines leveraging the Internet of Things (IoT) technology. The integration of IoT ensures real-time monitoring and data collection from different sensors installed on wind turbines, including infrared sensors, temperature sensors. This data is transmitted to a centralized monitoring system where advanced analytical techniques such as machine learning algorithms are employed for fault detection. The proposed approach offers several advantages over traditional methods, including early detection of faults, predictive maintenance scheduling, and reduced downtime. Furthermore, by utilizing IoT technology, remote monitoring and diagnostics can be performed, allowing for timely intervention and optimization of maintenance resources.The effectiveness and efficiency of our proposed approach is demonstrated through case studies and simulations, highlighting its capability to accurately detect various types of faults in wind turbines. Overall, this research contributes to enhancing the reliability and performance of wind energy systems, ultimately facilitating the transition towards a more sustainable energy future.

2. Empowering Indian Farmers: A Machine Learning Approach for Optimal Crop Selection and Sustainable Agriculture

   Ravi Charita, Kyasa Likhitha, Akella Samiksha, Baddam Arun, Raj Kumar Chanda, Pavan Kumar Pagadala

   Abstract

   Agriculture serves as the foundation of India's economy, being a pivotal occupation for a significant portion of Indian households. The agricultural sector, however, faces significant challenges due to shifting climate patterns and environmental shifts. However, many farmers persist in cultivating the same crops repeatedly due to a lack of knowledge about soil conditions. This practice leads to soil acidification and the erosion of the topsoil layer. To combat these challenges, we have developed a Machine Learning model tailored for farmers. Our model offers invaluable assistance by recommending the optimal crop choices based on prevailing weather conditions and soil health. Through our model, farmers gain insights into a variety of crops to cultivate, enhancing production, boosting profits, and mitigating soil pollution. The predictive capability of this model empowers farmers to anticipate crop yields before embarking on agricultural cultivation and offers farmer factors such as the means to proactively plan their agricultural activities. Lever-aging machine learning, this crop prediction model considers soil conditions, weather parameters, and historical crop data to make informed forecasts about suitable crops for the present circumstances. The model's training relies on historical crop data and pertinent parameters, encompassing water quality and soil conditions. The accuracy of our model is determined through rigorous training and testing, predominantly utilizing our dataset. Importantly, this tool is designed with the understanding that many farmers lack formal education and have limited knowledge about soil and weather conditions, and helps in ensuring better crop yields and increased profits for all.

3. Empowering Social Media Engagement: A Web Application for Analyzing and Categorizing Digital Content

   Rokkam Vivek Vardhan Reddy, Vidiyala Abhiram, Alwala Raghavendra Goud, M. V. S. Sai Teja, Sree Lakshmi Pinapatruni, Pavan Kumar Pagadala

   Abstract

   Social media is a fundamental aspect of modern society, offering a window into the values, conventions, and behaviors that shape our daily lives. By studying social media, we gain insights into how these platforms influence public opinion, social norms, and individual behavior. This research focuses on empowering users to make informed decisions about which social media sources they choose to follow. To facilitate this, we have developed a web application that allows users to monitor and categorize content from specific Facebook, Instagram, Snapchat, and YouTube accounts. With user-defined taxonomies, this application enables users to organize and analyze posts according to their interests and preferences. By providing this level of customization, the application supports users in navigating the vast amount of information on social media, helping them focus on content that aligns with their values and objectives. Ultimately, this research underscores the importance of understanding social media’s role in shaping contemporary society and provides tools that allow users to engage more thoughtfully with the digital content that permeates their lives.

4. Synergizing Literature Insights and Deep Learning for Effective Skin Cancer Detection and Classification

   Radhika Takkella, Pavan Kumar Pagadala

   Abstract

   This study integrates a comprehensive literature survey with the development and preliminary evaluation of a deep learning methodology for melanoma skin cancer detection. Our approach includes collecting a diverse dataset of skin cancer images, which undergoes rigorous preprocessing to enhance image quality essential for effective model training using TensorFlow and Keras. The models, built on advanced neural network architectures, demonstrate initial promise in accuracy, precision, recall, and F1-score metrics, underscoring potential improvements. We also prepare for the integration of this system into a web-based application designed to aid dermatologists in diagnosis, highlighting the synergy between detailed data analysis and practical medical application. This research underscores the critical role of precise data handling and advanced modeling techniques in enhancing diagnostic processes for skin cancer.

5. Artificial Intelligence Reshaping Human Resource Management

   E. Pranavi, Smitha Mahindrakar, T. Malathi Latha, V. Vijaya Lakshmi, J. Mamatha

   Abstract

   This Paper deals with transformative impact of AI on human resources (HR), aiming to shed light on its multifaceted roles, implications, emerging trends, and cutting-edge tools. With a focus on addressing key objectives, including studying the use of AI in HR functions, assessing its influence and identifying prevailing trends and tools, this paper provide a comprehensive overview of the evolving intersection between artificial Intelligence and HR. Through critical analysis and exploration of real-world case studies, this paper elucidates the potential benefits of Artificial Intelligence adoption in Human Resource processes, such as recruitment, talent management, and workforce analytics, while also considering the associated challenges and ethical considerations. By synthesizing current research and industry insights, this paper offers valuable insights into how organizations can navigate the AI-driven transformation of HR practices to enhance operational efficiency, decision-making, and strategic workforce management.

6. Ensemble Model for Exploratory Data Analysis and Prediction of Cardiomyopathy

   V. Kakulapati, J. Poornima, Y. Srinidhi, D. Greeshma

   Abstract

   This study intends to develop a dependable forecasting methodology for detecting cardiomyopathy via ML (machine learning) techniques such as Random Forests (RF), Decision Trees (DT), and K-Nearest Neighbours (KNN), hence improving prediction accuracy through the integration of different algorithms. EDA (Exploratory data analysis), which includes pre-processing procedures such as feature normalization and encoding categorical variables, helps in understanding the features of the dataset. We are using random data samples to train each classifier, and we'll evaluate their performance using metrics like confusion matrices and accuracy. A voting classifier is used to create the ensemble model, which combines predictions from different classifiers to maximize their combined intelligence and improve overall predictive accuracy. Finally, the ensemble model will predict cardiomyopathy in newly collected patient data, enabling early detection and personalized treatment planning. By incorporating advanced machine learning techniques, this study improves the way doctors make decisions and diagnose heart problems, leading to better treatment outcomes in cardiovascular medicine.

7. Revolutionizing Industry 5.0: The Role of Artificial Intelligence and Machine Learning in Manufacturing Techniques

   V. Vijaya Lakshmi, K. Syamala Devi

   Abstract

   This research paper examines the transformative impact of Artificial Intelligence (AI) and Machine Learning (ML) in advancing manufacturing techniques within the context of Industry 5.0 in India. The study aims to identify the adoption patterns of AI and ML technologies across various manufacturing sectors and to explore the factors influencing their successful implementation. Employing a mixed-methods approach, data were collected through surveys targeting middle to senior management across major industrial hubs including Mumbai, Pune, and Bangalore. The primary data analysis utilized clustering analysis to segment firms based on their AI and ML adoption levels and regression analysis to determine the factors affecting the success of these technologies.

   Key findings indicate distinct adoption patterns: Early Adopters, Mid-level Adopters, and Late Adopters, with varying degrees of integration and operational impact. Significant predictors of successful AI and ML implementations include budget allocation, technological infrastructure, and training and development. The results also reveal substantial regional disparities in adoption rates, with Bangalore leading in technology integration. Challenges identified include high initial costs, skill shortages, and technological complexity.

   The implications of this study are significant for policymakers and business leaders, suggesting that strategic investments in infrastructure, education, and skill development are crucial for maximizing the potential of AI and ML in manufacturing. The findings provide a roadmap for enhancing technological integration that could lead to increased competitiveness and innovation in the Indian manufacturing sector.

8. Vehicular Cloud Forming and Task Scheduling for Energy-Efficient Cooperative Computing

   L. Smitha, P. Yasaswini Reddy

   Abstract

   A vehicle-to-vehicle (V2V) network of automobiles that carries out cooperative computing is known as a vehicular cloud (VC). Most of the current research on vehicular cloud computing (VCC) relies on edge or cloud servers rather than virtual clouds (VCs). Nevertheless, without the assistance of edges or cloud servers, automobiles might carry out collaborative programs needing a significant quantity of computing by creating a Vehicular Ad-Hoc Network (VANET). One of the key challenges for VANET cooperative computing is how to deal with the frequent topology changes brought on by moving cars. Cooperative computing's benefits are limited by an unstable network architecture, which occasionally even causes it to cease operating. In this article, a vehicle-to-vehicle (V2V) communication-based cooperative computing approach is proposed. The suggested solution takes the distance into account while choosing which cars to collaborate on and delayed task offloading back as far as feasible for reliable and energy-efficient cooperative computation. The suggested approach performs better in terms of both energy efficiency & network stability than earlier static scheduling techniques.

9. Enhancing Parkinson's Disease Diagnosis Using Genetic Algorithms

   Sireesha Vikkurty, Nagaratna P. Hegde, Sriperambuduri Vinay Kumar, Kaligota Shireesha, Devireddy Rukvith Reddy

   Abstract

   Parkinson’s Disease presents a significant worldwide health challenge, impacting millions with its motor and non-motor symptoms. Timely detection and precise prediction are crucial for improving patient outcomes. In this project, we introduce an innovative strategy for Parkinson’s Disease risk prediction utilizing a Genetic Algorithm (GA) in conjunction with machine learning techniques. The GA optimizes the selection of genetic markers and pertinent features from an extensive dataset comprising demographic details, medical records, and clinical assessments linked to Parkinson’s Disease. Through iterative refinement, the GA identifies the most informative feature subset vital for predicting disease susceptibility. Different machine learning models are trained using the chosen features. The efficacy of each model is assessed and our proposed methodology demonstrates superior accuracy in predicting Parkinson’s Disease risk compared to existing approaches. By combining GA-based feature selection with machine learning models, our approach enables precise and effective Parkinson’s Disease prediction, facilitating early diagnosis and tailored therapeutic strategies. This study underscores the potential of genetic algorithms in enhancing predictive models for neurodegenerative conditions like Parkinson’s Disease.

10. K-Nearest Neighbors and Support Vector Machine for Optimal Content-Based Image Retrieval with Low-Level Feature Fusion

    M. Narayana, Manoranjan Dash, N. Mangala Gouri, Avadutha Rachana

    Abstract

    Using a technique called content-based image retrieval, images can be found and retrieved from image databases according to their visual content. The absence of thorough representation is a major problem because different features frequently only capture some facets of the visual information. The application of K-Nearest Neighbors and Support Vector Machines, two well-liked machine learning methods, for CBIR. To create a thorough feature vector for every image, the suggested method extracts low-level data like color histograms, texture descriptors, and form features. These feature vectors are used to independently train the KNN and SVM classifiers, creating strong models that can predict picture similarity. The goal is to obtain a more comprehensive representation of the image content by combining the strengths of individual low-level elements through the use of a fusion technique. The study assesses and contrasts the retrieval accuracy, computational efficiency, and scalability of KNN and SVM. In order to attain improved retrieval efficiency, our suggested system makes use of a unique framework that incorporates texture, color, and shape information. To enable precise retrieval of the needed photos, the system gathers a wealth of significant, reliable, and comprehensive information from the image database and saves them in the repository.

11. Sensorless Speed Detection in BLDC Motor Using Artificial Intelligence

    B. Deekshitha Reddy, Kunta Srikanth, B. Madhuri

    Abstract

    The precise estimation of motor speed is crucial for a wide array of industrial applications. In our project, we propose an innovative method for sensorless speed estimation of Brushless Direct Current (BLDC) motors, leveraging Artificial Intelligence (AI) techniques to circumvent the need for conventional sensor feedback. To ensure the accuracy and dependability of our findings, we have adopted a comprehensive approach in our study. Alongside the development of artificial neural network (ANN) models using Python, we have constructed a complementary mathematical model using MATLAB. The MATLAB model functions as an independent validation tool, allowing us to cross-validate the results obtained from the ANN models. By integrating the fundamental principles governing BLDC motor dynamics—such as electromechanical equations and voltage-current-speed relationships—we have constructed a detailed mathematical representation within MATLAB. The proposed model includes important characteristics including motor constants, electrical and mechanical qualities, and external torque influences to provide realistic modeling of real-world events. We investigate various operating situations using simulations made possible by this model in order to accurately predict motor speed. We compare the speed estimates produced by our MATLAB scripts with the results of our ANN models by doing simulations. Any differences between these outcomes are carefully examined to find any inconsistencies or areas that need improvement. This dual strategy not only confirms the accuracy of our speed estimation procedure but also provides insightful information on the advantages and disadvantages of mathematical and ANN-based modeling approaches. Our goal is to create a solid foundation for sensorless speed estimation in BLDC motors by means of extensive.

12. Vehicle Number Plate Detection and Recognition Using YOLOv7

    Uppula Yogeeshwar, Errolla Bhasker, V Nikesh, Saroja Kumar Rout, Nirmal Keshari Swain, Nilamadhab Mishra

    Abstract

    This paper presents an effective method for object detection in YOLOv7 (You Only Look Once version 7) for number plate recognition. The process entails precisely localising number plates and identifying automobiles in photos. With the help of transfer learning, YOLOv7 is trained on a unique dataset featuring a variety of number plate types and lighting situations. The system, which has been tested in difficult situations like occluded plates and complicated backgrounds, combines high accuracy with real-time processing. This methodical approach improves automatic vehicle identification systems, which in turn helps with parking management, law enforcement, and traffic monitoring, all of which lead to increased road safety and security.

13. Automatic Weapon/Defense System

    S. P. V. Subbarao, T. Ramaswamy, Supraja Sairi, Nithin Ramisetti, Saiteja Byri

    Abstract

    The “Automatic Weapon/Defense System” represents a state-of-the-art solution for threat detection and response in border areas and maritime environments. The project using a radar system excels in measuring distances and quickly identifying potential threats within a specified range. Once a threat is detected, an alert mechanism with a buzzer is activated to inform the system operator in real time. For a more affordable simulation of an expensive defense system, the project includes innovative components. An ultrasonic sensor serves as a cost-effective alternative to a radar system that effectively measures distances. HuskyLens takes care of threat tracking, constantly analyzing the movement of potential threats and ensuring accurate tracking. The warning system is triggered when threats exceed a specified range, approximately 40 cm in the simulation. The simulated firing sequence includes a manual control aspect where a system operator located nearby can trigger the firing mechanism after assessing potential damage. Weapon simulation uses a DC motor and threat tracking is facilitated by pan-tilt servos that provide stability during motion analysis. This cost-effective approach mitigates the high costs typically associated with implementing such defense systems. The project's innovative use of affordable sensors and simulation components demonstrates its potential for practical applications in real-world scenarios, promising increased safety and rapid response capabilities in critical areas. In order to simulate the project idea, it involves large and expensive machinery, so it has been scaled down to short distances. This approach allows to use more cost-effective components while maintaining the essence of the original concept.

14. Voice Responsive and Vision Guided Assistance Robot Using Arduino

    M. Vijaya Lakshmi, G. Swathi, K. Chandana, A. Sri Chandraja, K. Bhavana

    Abstract

    In the fast-changing technological world, the incorporation of smart gadgets and robotics into daily life has become common. Through this project we are taking a step towards tackling the demand for intuitive assistance applications by creating a low-cost wheeled robot equipped with voice response and vision-guided capabilities using Arduino technology. This robot is a dependable assistant in a variety of settings, smoothly blending into different situations and meeting a wide range of support requirements. This robot provides a natural manner of interaction by utilizing speech recognition and vision guiding, boosting the user experience and autonomy. Real-world demonstrations demonstrate the robot's capacity to overcome obstacles, find things, and give timely help. The suggested approach offers a huge step forward in human-robot interaction, with a focus on developing solutions that empower individuals and improve their quality of life via seamless and intuitive support applications.

15. Crime Pattern Recognition

    K. Sridevi, K. Sai Pavani, P. Vennela Reddy, D. Veda Smriti

    Abstract

    Crime is a widespread social problem that crosses cultural, socio eco- nomic, and local communities. Criminal activity has a significant effect on families, and society at large. Projects pertaining to crime are necessary be- because it is urgently necessary to address and lessen the detrimental effects of criminal activity. A significant component of crime analysis is data mining. Machine learning and data mining are essential tools for reducing crime because giving legislators and law enforcement organizations strong tools to evaluate enormous volumes of data, spot trends, and anticipate future criminal activity. Supervised, unsupervised, and reinforcement learning algorithms are examples of machine learning algorithms that further improve the efficacy of crime reduction. Machine learning algorithms can estimate the likelihood of criminal episodes occurring in particular places and time periods by training predictive models on previous crime data. Because predictive capacity, law enforcement organizations are able to take preventative action before crimes occur by stepping up patrols or focusing on specific areas. Algorithms that use machine learning techniques can help identify people or groups who are more likely to commit crimes. In this work, we identified crime analysis using linear regression and support vector machines (SVM). These algorithms are used to forecast the rate of crime and the number of crimes that will happen in the near future. Law enforcement agencies can utilize this information to reduce crime and allocate resources more effectively.

16. Visual Lane Tracking and Curvature Measurement System

    Nagaratna P. Hegde, Sireesha Vikkurty, Sriperambuduri Vinay Kumar, Maheshwar Reddy Somu, Pranav Jallapalli

    Abstract

    Detection of Lane departure assumes a critical role in bolstering the safety features of Advanced Driver Assistive Systems, profoundly influencing active and secure driving experiences. This project introduces an extensive lane detection methodology grounded in computer vision techniques, leveraging video streams captured by a roof-mounted camera atop a vehicle. The intricate process involves meticulous correction of camera distortion, the strategic application of HLS and Sobel operation with threshold to accentuate lane lines in a binary image, and subsequent perspective transforms that morph the resultant frame into a bird’s-eye view. Identification of the lane lines are performed based on a sliding window search and next discerned through the fitting of second-degree polynomials. The lane identification process includes computations regarding lane centre deviation and lane curve. The identified lane boundaries are re-projected onto the input image to facilitate the calculation of lane curve and vehicle position. Python programming language used for implementing this technique and OpenCV for image processing. Across a spectrum of challenging lighting conditions it performs consistently and accurately in detecting lane lines.

17. Intelligent Detection of Weeds in Crops Using Deep Learning Approach

    K. Ragasritha, N. Navatha, Hemanth Surya Sai Sunkara, B. Shailesh Chowdary, Madala Sreshta, Rajitha Ala

    Abstract

    Agriculture is important to society and needs to be planned, researched, and carried out. Investigating novel approaches, cutting-edge techniques, and potential accelerators is crucial. LATEX. With some technologies that improve search quality, the farmer can lessen the amount of labour. In the realm of convolution neural networks, it is crucial to determine the growth estimation of marijuana utilizing deep learning technology. This review study lists the various weed species that are detrimental to crops. This review paper summarizes the advancements in artificial intelligence and image processing approaches for the detection of weed and its classification. It uses the most recent techniques available. The four steps which are involved in weed detection and classification are (i) Preprocessing, (ii).Segmentation, (iii).Feature extraction, (iv).Classification which were specifically described in depth. Lastly, the difficulties and solutions that researchers had offered for classifying and identifying weeds in the field were covered.

18. A Real-Time Full Stack Chat Application Using AWS and NextJS

    Sri Harshini, Salma Saher, Koustubha Madhavi Balakrishnan, V. Mohan

    Abstract

    Modern chat applications require a variety of strong functionalities like file storage, real-time updates, and the ability to fetch data seamlessly on both client and server sides. However, these requirements were traditionally met by either integrating multiple third-party services from different vendors or dedicating considerable resources to customizing the solution. Nevertheless, this method has often resulted in late time-to-market and higher vulnerability to failures. We therefore propose an AWS cloud-based real-time chat application. Our proposed model consists of the latest AWS cloud solutions that include Amplify, Cognito, DynamoDB, and AppSync services for building the chat application. The integrated AWS services encompass all functionalities needed for a real-time chat app. This study therefore contributes to how the operations of a cloud can be employed together with each other to produce an entire live chat app without using external sources. The results are convincing and show that the intended objective was achieved by our framework proposal.