Volume 17, Issue 2, April 2026

Effect of Microstructure on the Tensile Strength and Impact Toughness of Gray Cast Iron

Jimoh S. O., Samuel O. Igbudu, Ediale Ameya Lucky

Department of Industrial and Production Engineering, Faculty of Engineering and Technology, Ambrose Alli University, Ekpoma

Abstract- Ferro-silicon of different percentage (1.00w%, 1.50w%, 2.00w%, 2.50w% and 3.00w%) in the chemical composition, together with constant amount of cast iron scraps and graphite 1.0kg each, were used to produce gray cast irons samples. A mixture of 80% white silica sand, 10% bentonite as binding agent 2% coal dust as carboneous material to enhance permeability and 8% water to activate the clay and to enhance mod-ability and flow-ability of the sand, were used to prepare the mould. The mould were molded in moulding boxes and carbon (iv) oxide was conducted to pass through the mould in order to eliminate moisture and to enhance the hardness of the mould. The charge materials were introduced into a rotary furnace and heated to 13000C to produce melt. 1.00% of Ferro-silicon was added to the mixture and the furnace was tilted to allow the melt pass through the out let into a preheated ladle and then poured into the prepared mould of the same dimensions (20mm diameter and 25mm long). The melt was repeated for more four times for different percentage of ferrous silicon (1.50w%, 2.00w%, 2.50w% and 3.00w% in the chemical composition. The melt was allowed to cool and solidify for 24 hours in order to obtain solidified cast samples. Polishing machine, emery paper of different grits and etchant solution were used to prepare the surfaces of the cast sample for microstructural examination. The sizes and orientation of the graphite flakes in the ferrite matrix through observation differ from one sample to another as the amount of Ferro-silicon varied. The samples were also subjected to mechanical testing such as tensile strength and impact toughness, using Intron Universal and impact testing machine respectively. The result shown that the tensile strength of the test samples decreases with increase in the percentage of Ferro-silicon, however, the impact toughness increases with increase in the percentage of Ferrous-silicon.

Keywords- Cast Samples, Chemical Composition, Ferro-Silicon, Ferrite Matrix, Graphite Flakes, Impact Toughness and Tensile Strength

Download full paper PDF format (Page: 1-6)

Identification of Rice Plant Disease Detection via IoT-enabled Intelligent Irrigation System with Advanced Machine Learning Techniques

Mujtaba Kamal Pasha, Ramsha Khalid and Saad Rehman Babary

Department of Computer Science & Engineering, University of Engineering and Technology, Lahore

Abstract- Rice is one of the most important groups of crops consumed by the population of Earth since several million rely on it for nutrition. However, these plants are still susceptible to a variety of diseases that adversely affect their yield and quality. This research presents the use of machine learning-based algorithms in the early detection and management of rice plant disease. A dataset is prepared with the collected details regarding traits of disease symptoms and designs it for supervised and unsupervised learning models. The classification of diseases is done through deep learning models while clustering techniques help in the interpretation of the path of disease propagation. Furthermore, integration with the camera sensor is studied for real-time monitoring and better predictive work. The disease detection used here is based on CNNs with EfficientNetB4, YOLOv8, InceptionV3, and ResNet50. The given model for training is composed of twelve disease classes and healthy rice plants. The results yielded an experimental result that EfficientNetB4 training yields 89.95% accuracy, YOLOv8 gives 84.34% accuracy, InceptionV3 achieves 82.85%, and ResNet50 achieves 36.16%. Moreover, rice-leaf disease recognition is also included, wherein ResNet-34 with LSTM gives 97.7% accuracy outperforming other CNN and SVM models. The predictive modeling based on ongoing sensor data makes the disease detection process easier and more scalable. The research findings bolster the precision agriculture being AI-driven, enhancing crop health and bringing about better sustainable agricultural practices.

Keywords- Artificial Neural Network, Convolutional Neural Network, EfficientNetB3, EfficientNetB4, InceptionV3, Machine Learning, Natural Language Processing, Random Forest Tree, ResNet50, Sentiment Analysis, Support Vector Machine and YOLOv8

Download full paper PDF format (Page: 7-14)

IoT-Based Smart Monitoring and Forecasting for Pallet Production: A Review

Noorulain

Department of Data Science, UET (Main Campus)

Abstract- Modern manufacturing systems increasingly rely on Internet of Things (IoT) technologies to enable real-time monitoring and intelligent decision-making. IoT-based systems generate large volumes of time-series data from industrial processes, which can be leveraged for predictive analytics. However, most existing manufacturing systems still rely on static or historical estimation methods for production planning, leading to inefficiencies and suboptimal resource utilization. This paper presents a review of IoT-based monitoring systems and time-series forecasting techniques used in pallet production environments. The reviewed studies utilize real-time sensor data combined with forecasting models such as ARIMA and Long Short-Term Memory (LSTM) networks to predict short-term production output. The surveys framework also introduces a pallet ranking mechanism for prioritizing production based on demand and operational conditions. The reviewed literature indicates that integrating IoT monitoring with advanced forecasting models can improve prediction accuracy, optimize production planning, and support data-driven decision-making. This review identifies the need for unified IoT-driven monitoring and forecasting frameworks in smart manufacturing system.

Keywords- IoT, Time-Series Forecasting, LSTM, ARIMA, Industry 4.0 and Pallet Production

Download full paper PDF format (Page: 15-24)

DistilBERT-Based Bug Triage with Attention Mechanisms for Enhanced Developer Assignment

Syed Faizan Alam Zaidi

Departmrnt of Computer Science, India

Abstract- Abstract— Effective bug triaging plays a critical role in efficient software maintenance. Conventional manual bug triaging is time-consuming, error-prone, and frequently misidentifies developers who should be assigned to fix bugs, thereby delaying the bug resolution process and reducing software quality. Despite advances in automated bug triaging, existing approaches face key challenges, including limited semantic understanding of bug reports, poor scalability to large repositories, and suboptimal performance in Top-1 accuracy despite reasonable Top-k results. In this paper, we present a bug triage model which utilizes transformer-based NLP model and attention-based classification head to accurately assign bugs to the right developers. Our model uses DistilBERT to obtain contextualized representations of the bugs' descriptions, and experiments with two classification methods: a linear attention-based classification head and a BiLSTM classifier with attention. The proposed approach is tested using two massive publicly available bug triage datasets from Mozilla Firefox project. Our findings reveal that an attention-based classifier outperforms the BiLSTM-based one and achieves better Top-1 accuracy up to 40% and Top-10 accuracy of up to 69%. These findings highlight the effectiveness of combining transformer-based semantic representations with attention-driven classification to enhance both Top-1 precision and Top-k robustness.

Keywords-Bug Triage, Developer assignment, DistilBERT, Attention Mechanism, Deep Learning and Software Maintenance

Download full paper PDF format (Page: 25-36)

Abstract- Abstract—

Keywords-

Download full paper PDF format (Page: 37-45)