×
The submission system is temporarily under maintenance. Please send your manuscripts to
Go to Editorial ManagerSoil classification is a significant tool for the interpretation of soil properties and the selection of the appropriate parameters to express soil types in geotechnical engineering applications. In this paper, a comparative analysis on the prediction of soil classification using different soil parameters via Neural Network, Decision Tree, and Ensemble classifier algorithms in the machine learning tool box of MATLAB is conducted. The data is classified into 14 sets of input parameters according to the parameters of the Unified Soil Classification System (USCS), which is a standard system in practice for classifying soils based on their physical properties and their engineering behavior. The input parameters of the USCS, Passing No. 4 Sieve, Passing No. 200 Sieve, Gravel, Sand, Fines, Liquid Limit, Plastic Limit and Plasticity Index, are specified within the certain ranges to generate a synthetic data base including 478 data for the purpose of training and testing. Five Neural Network models (Medium, Trilayered, Bilayered, Narrow and Wide), three Decision Tree models (Fine, Medium, Coarse) and three Ensemble models (Boosted Trees, Bagged Trees, RUSBoosted Trees) are tested for the different test split ratios (10, 20, and 30). The results show that the Neural Network models proposed in this study have the highest prediction capability, with a classification accuracy of up to 100 % for some variable sets, and can classify soils automatically and precisely based on the soil index properties. Among the individual models, WNN and NNN generally showed the best performance for all variable sets. The Ensemble models presented high prediction capability with respect to generalization. Among the DT models, although they are fast, they presented relatively lower accuracy than the other models, especially the Coarse Tree model. This study can be used as a reference for the future development of automated soil classification systems and can offer practical suggestions for the selection of machine learning models for geotechnical engineering problems.
Plants and agriculture are important in Iraq and the world because they are among the essential basics of life; their importance lies in several fields, such as industry and food. Plant diseases are the first direct influence on plant production and the Iraqi economy. The primary contribution of this work is developing an efficient early warning system for tomato plant diseases based on readily available environmental data, demonstrating the usefulness of machine learning methods in real agricultural environments. This research investigates the use of artificial intelligence (AI) for the early prediction of tomato diseases in Iraqi agriculture, based on temperature and humidity data collected from Salah al-Din Governorate. Two major diseases were studied: Tomato Yellow Leaf Curl Virus (TYLCV) and late blight. The data were pre-processed and used to train predictive models, including linear regression and Random Forest Regressor (RFR). Results show that RFR outperformed linear regression, achieving a lower Root Mean Square Error (RMSE) of 0.053852 and a Mean Absolute Error (MAE) of 0.45000, indicating its superior accuracy in predicting disease occurrences.
AISI 4330 Low-alloy steel is good material for advanced application because of its properties including strength and longevity. However, performance may be modified with heat treatment procedures, include quenching and tempering. These processes can create residual stresses and retained austenite (RA), which have an effect on the metal's application. these factors influence fatigue life, dimensional stability, and fracture toughness of engineered components. uncontrolled residual stresses can reduce fatigue strength by up to 30%, while optimal retained austenite content (e.g., 5-10%) can enhance damage tolerance. This study focuses on residual stresses and retained austenite measurement in AISI 4330 low-alloy steel after heat treatment. including experimental and simulation methods. The review summarizes many scientific studies published between 2019 and 2024 and shows some main challenges. One challenge is the difference between experimental results (for example, from X-ray diffraction (XRD) and neutron (diffraction) and simulation results (especially using ANSYS software). Another challenge is that different methods for measuring retained austenite can give different results, which can change how we understand the steel's properties. The review also explains new progress in modeling heat treatment. This includes adding phase transformation models to finite element simulations. Future efforts should combine multiscale simulation, characterization, and machine learning to achieve predictive control over these properties in manufacturing.
Image segmentation is the process of automatically dividing an image into distinct, meaningful, and non-overlapping regions. The quality of the segmentation process determines the efficiency of other image processing tasks. Analyzing microstructural images is crucial since the mechanical properties are strongly dependent on the microstructural phases’ statistics. These images are considered one of the most difficult and challenging images to deal with due to their special characteristics, such as the convergence in pixels intensity values, overlapping in colors, boundaries and textures in phase regions, infinite shapes of grains and colonies, etc. As there is no generic technique suitable to be used with all microstructures, this work reviews techniques that have been effectively used and recommended to be employed in metallurgical research, with a brief description of their principles, advantages, and disadvantages, and discusses their applicability. The major aim of this work is to spare time and effort searching for and experimenting with all the available methods for future researchers.
The accurate prediction of machinery faults is considered an effective strategy to increase the operation life of machines, ensure smooth operation, and provide a safe environment. Accordingly, the demands on predictive tools such as machine learning to detect machinery faults before catastrophic failure occurs has increased rapidly. In this research, a diagnosis algorithm based on using a 2D color-coded map as the input to a deep artificial neural network is proposed. These maps are called RDEgram after the processing of vibrational signals based on reverse dispersion entropy (RDE) method. The effectiveness of the proposed algorithm is investigated by testing its capability to detect different faults located at different locations on ball bearings under constant speed conditions. First, the squared envelope signal is extracted by applying the short time Fourier transform to vibration signal. Then, the RDE is used to process the squared envelope to detect the range of frequencies at which the transients occur. The RDEgram color-coded map is used to represent the RDE values as a function of frequency and frequency resolution. The maps from different fault features are collected to form the diagnostic patterns. Finally, a pretrained convolutional neural network (CNN) is applied to learn the feature pattern and diagnose the bearing faults. The CNN is trained using fixed- speed data and then it is applied to diagnose faults in the test data recorded at the same speed. The prediction method adopted in the current research shows a 100% level of accuracy for predicting two types of faults (pit and slot) located at various positions a ball bearing (KOYO 1205 C3 type) running at two constant speeds (25 and 30 Hz).