Publications

Seismic deformation assessments are an ongoing issue in the design, monitoring and construction of earth dams. The need for new advanced methods to model their seismic behavior and to evaluate the resulting deformations is justified by the uncertainties surrounding conventional methods, mainly, with liquefaction phenomena. In this respect, the present study focuses on the prediction of relative crest settlement of embankment dams under variant earthquake loading (Δh_EQ/H). For this purpose, Back-Propagation Neural Network (BPNN) and Multivariate Adaptive Regression Splines (MARS) models were developed to predict (Δh_EQ/H). Two different databases of historically documented earthquake cases are collected for model development and comparative performance between model predictions. The first contains 151 observations of liquefied and non-liquefied cases, while the second contains only 109 non-liquefied cases. The obtained results indicated that both technics could be used as reliable tools to predict the earthquake-related crest settlement in embankment dams. Also, MARS was selected as the most successful prediction tool.

In this work, a modified perturbs and observes (P&O) technique is used in the hybrid power generation system with power management. There are several algorithms for extracting the maximum power point (MPP) provided from the PV generator; P & O algorithm has a good performance, ease of implementation by analog and digital electronics. However, this algorithm has disadvantages because it oscillates at the point of maximum power and has a relatively long convergence time; therefore, a modification is made to the P & O algorithm in order to reduce setup time and oscillations in the MPP. The proposed system allows optimal use of the photovoltaic (PV) system and the DFIG wind proves its efficiency under variable load conditions. The model is implemented in the MATLAB / Simulink platform.

This paper proposes an optimum design of a diesel/PV/wind/battery hybrid renewable energy system (HRES) for rural electrification in a remote district in Tamanrasset, Algeria. In this study, a particle swarm optimization algorithm (PSO) has been proposed to solve a multi-objective optimization problem, which was created by carrying out simultaneously, the cost of energy (COE) minimization while maximizing the reliability of power supply described as the loss of power supply probability (LPSP) and a renewable fraction (RF). The simulation results show that the PV/WT/DG/BT is the best economic configuration with a reasonable annual cost of the optimal system (ACS) which is about 7798.71 $ and the COE equal to 0.79 $/kWh for an LPSP = 0.01%, where the ten households are 0.99 % satisfied by renewable energy sources.

In this paper, we present a system for handwriting digit recognition using different invariant features extraction and multiple classifiers. In the feature extraction we use four types: cavities, Zernike moments, Hu moments, Histogram of Gradient (HOG). Firstly, the features are used independently by five classifiers: K-nearest neighbor (KNN), Support Vector Machines (SVM) one versus one, SVM one versus all, Decision Tree, MLP. Then to achieve the best possible classification performance in terms of recognition rate, three methods of classifiers Combination rule employed: majority vote, Borda count and maximum rule. Experiments are performed on the well-known MNIST database of handwritten digits. The results demonstrated that the combination of KNN using HOG features with SVMOVA using Zernike moments by Borda count rule have considered to be good based on a geometric transformation invariance.

This paper proposes a new approach of multi-spectral image compression based on the combination of the particle swarm optimization (PSO) and the discrete wavelet transforms (DWT). In the first stage, the PSO is used to reduce the redundancies in the spectral domain. In fact, the PSO transforms a given multispectral image to optimize the energy in the first band. Despite to the complexity of this kind of approach, the transformed multispectral image is easily computed by multiplying a de-correlation matrix and the input multispectral image. The de-correlation matrix is estimated via a PSO evolution derived by a defined fitness function. In the second stage, the compressed data, related to the input multispectral image, is computed from the transformed multispectral image using an efficient 2D-DWT based algorithm. In addition to this compression approach, the original multispectral image can be recovered using a decompression algorithm. Experimental results show the validity of our proposed approach. These significant results are evaluated according to Peak signal-to-noise ratio (PSNR), compression ratio (CR) and bits per pixel (bpp) metrics.

The present study investigates the spectral analysis for natural convection in a tilted rectangular cavity, lled with high Prandtl oil ”Pr =880” by the code CFD. A constant vertical temperature gradient has been performed by subjecting the horizontal walls to constant temperatures Th and Tc; respectively. Other walls are adiabatic except the left small sidewall is differentially heating with temperature TA creating the horizontal temperature gradient. The results are presented for different values of lateral heating and inclination angle. The spectral analysis is used to identify and show effects on the original oscillation of the natural convection by the various investigated parameters (TA and θ)