S.L.N College of Engineering

The poultry business has a considerable impact on the food manufacturing sector. Poultry hens are in high demand, but there are growing worries over their quality in many parts of the world. Quality control in the poultry sector helps to ensure a steady supply of eggs and meat. The industry’s stakeholders benefit from recent technical developments in the tracking and monitoring of bird health. To address this issue, the authors present a system that uses a combination of a (LSTM-RF) to extract features from chicken images in an effort to boost identification accuracy. In order to test the efficacy of the suggested LSTM-RF despite the imbalanced dataset, many data improvement strategies were developed. When applied to chicken age detection, the suggested model increases accuracy to 95%, as shown by the final testing results. In addition to the aforementioned mobile-ready detection software, this study also presents the design of a comprehensive image-acquisition system for aviaries. Also, a more accurate and high-performing classification approach is provided by comparing the accuracy of several classification models. The main goalmouth of the investigate is to deliver a prognostic service framework based on the Industrial Internet of Things (IoT) that can more precisely categorise fowl hens in real time.

Vehicular Ad-Hoc Networks, also referred as VANETs, have emerged as an interesting area of research as a result of ever significant increase in the number of automobiles on roads. Built - in safety smart vehicular traffic infrastructure protects both passengers and drivers, but due to its dynamic nature, real-time implementation is difficult. They lay the groundwork for the creation of intelligent transportation systems (IPS) and frameworks, which allow road entities to communicate with one another and build new applications and services with the goal of improving both the driving experience and overall road safety. The demanding features of VANETs make it difficult to establish security measures, resulting in gaps that attackers could exploit. This research provides smart structured protective systems for VANETs that use machine learning (ML) based algorithms. The enhanced ML algorithms improves attack detection, protecting data inter-communications between various sources and destinations, and ensuring strong anonymity, authentication, and privacy. The proposed system trains and tests its machine learning algorithms on publicly available datasets of vehicle communications. As a result, the outcomes are repeatable and verifiable. The machine learning-based security system can detect attacks while maintaining low False Positive Rate values (FPR). The findings also suggest that the framework may benefit from employing a variety of algorithms present at various hierarchical levels, selecting algorithms with high performance and focus at the cost of preciseness in lower levels and additionally sophisticated, detailed, and accurate algorithms present in top levels.

Vehicular Adhoc Networks (VANETs) have been more popular over the last decade. Vehicles now have sophisticated sensors that accomplish anything from automatically monitoring lanes to preventing front-end crashes to providing semi-autonomous driving to suggesting lane changes. Because of technological advancements, VANETs can now provide a broad range of useful services to drivers and passengers, including making the road trip more pleasant and safer. The data on traffic accidents and safety that is sent through an Adhoc network in a vehicle must be delivered unaltered. It is possible for hostile vehicles to hack into and intercept transmissions from vehicle Adhoc networks. Vehicle Adhoc networks can experience disruptions in both their performance and security. According to the research provided here, a three-tiered trust management system is a viable strategy for dealing with hostile vehicles in vehicle Adhoc networks. The proposed system begins with a trust rating for each vehicle in the Adhoc network, calculated from the vehicle's processing time, packet loss rate, and prior behaviour. Pick out certain vehicles to keep an eye on the flow of traffic around them so we know which ones can be trusted. The second-layer defenders are safe from malicious activity credits to past actions. The third layer of security safeguards the information needed to determine trustworthiness. Suggested method remains effective for verifying rogue VANET automobiles. When malevolent vehicles are added to a vehicle Adhoc network, data-packet transmitting rate increases, while source-to-destination latency decreases.

A malicious URL is one that was made specifically to attack through spam or fraud. Due to the billions of dollars that are compromised, malicious URLs pose a severe threat to security software. Finding secure and phishing links is therefore crucial. Therefore, machine learning is quite helpful for resolving security-related challenges. In this study, we use about 5 lakh URLs that were retrieved from the Kaggle dataset. We are utilizing three NLP approaches, including the count vectorizer, hash vectorizer, and TF-IDF vectorizer. Six machine learning classifiers, including the decision tree, random forest, K-NN, NB, SVM, and logistic regression, were used in conjunction with all these techniques. The highest accuracy results of 98.2 percent are produced by random forest. To determine whether or not the URL supplied is malicious, we built a web app using Flask.

Fuel properties are essential components in the performance and emission of internal combustion engines. To measure density, viscosity and other properties of methyl esters and two methyl esters blends in the temperature range of 25°C-95°C. Experimental investigations is carried out using standard equipments and methodology. Test fuels considered are Pongamia, Jatropha and Simarouba biodiesels and blends of two biodiesels in different volume fraction of 0% to 100%. Density of Pongamia-Jatropha and Pongamia-Simarouba biodiesel blends increases linearly with increase in volume fraction of Pongamia methyl esters at all the temperatures. Density of Jatropha-Simarouba biodiesel blends increases with increase in volume percentage of Jatropha in the blends. The kinematic viscosities of blends are found to reduce almost logarithmically with increase in temperature. Viscosity of the blends lies between the constituent fuels. The measured value of density and viscosity is correlated as function of blend percentage and temperature through an empirical relation. The correlations developed are unique and are model equations. The maximum % density variation between the experimental results and estimated results is 0.4%, where as percentage viscosity variation between experimental results and estimated results from correlation is 5%.