Aref Maalej

A Genetic Designed Beta Basis Function Neural Network for Approximating Multi-Variables Functions

We propose in this paper a new genetic algorithm for Beta basis function neural networks (BBFNN). The properties of this ’genetic algorithm are the representation used and the ability to obtain the optimal structure of the BBFNN for approximating a multi-variable function.

The design of beta basis function neural network and beta fuzzy systems by a hierarchical genetic algorithm

We propose an evolutionary method for the design of beta basis function neural networks (BBFNN) and of beta fuzzy systems (BFS). Classical training algorithms start with a predetermined network structure for neural networks and with a predetermined number of fuzzy rules for fuzzy systems. Generally speaking both the neural network and the fuzzy systems are either insufficient or overcomplicated. This paper describes a hierarchical genetic learning model of the BBFNN and the BFS. In order to examine the performance of the proposed algorithm, it is used for functional approximation problem for the case of BBFNN and for the identification of an induction machine fuzzy plant model for the case of BFS. The results obtained have been encouraging.

A hierarchical genetic algorithm for the design of beta basis function neural network

We propose an evolutionary neural network-training algorithm for beta basis function neural networks (BBFNN). Classic training algorithms for neural networks start with a predetermined network structure. Generally the network resulting from learning applied to a predetermined architecture is either insufficient or over-complicated. This paper describes a hierarchical genetic learning model of the BBFNN. In order to examine the performance of the proposed algorithm, they were used for the approximation problems. The results obtained are very satisfactory with respect to the relative error.

Study the effect of tool geometry and operational conditions on mouldboard plough forces and energy requirement

Experimental validation of FEM results in part 1 using soil bin.The effect of the depth and the speed on tillage forces.The influence of the cutting angle (α) and the lifting angle (β) on tillage forces.The influence of the cutting angles (α and β) on the soil loosening and inversion.The influence of the depth and speed on the soil loosening and inversion. Forces acting on tillage tools directly affect vehicle fuel consumption of tillage operations. Both tool forces and soil disturbance are a function of tillage tool type, tool geometry, and operational conditions. A soil bin experiment in a sandy loam soil was conducted to validate the results obtained from the finite element method (FEM) simulation of the interaction between soil and a mouldboard plough, carried out in part 1 of this study. An octagonal load cell was used to measure the draught and vertical forces. A special support was manufactured to modify the cutting angle (alpha) and the lifting angle (beta). Tillage forces and soil disturbance were measured for different speed (0.5, 1, 1.5, and 2m/s), depth (100, 150, 200, and 250mm), cutting (30?, 45?, 60?, and 75?), and lifting (25?, 40?, and 55?) angles, and were presented for the same soil conditions of those considered in the FEM. The soil disturbance including the width and surface area of the cut soil, and the height, width, and the surface area of the soil inversion were measured with a laser scanner. Results showed that both the FEM calculations and the soil bin measurement presented the same tendency for the variation of draught and vertical forces with speed, depth, cutting, and lifting angles. The maximum error recorded between the measured and the FEM results was 33.8%. It was found that the draught and vertical forces increased linearly with speed, whereas a second order polynomial and linear relationships were established with depth, respectively. Draught increased linearly with both the cutting and lifting angles, whereas, the vertical force decreased linearly with these angles. The study of the soil disturbance showed that the operating conditions (speed, depth, and cutting angles) of the mouldboard plough had an important effect on the quality of the tillage (soil loosening and inversion). Both the FEM and the soil bin tests showed that at a working speed of 1m/s and a depth of 150mm with lower lifting and cutting angles of 25? and 45?, respectively, provide the best combination for lowering energy consumption. The soil bin tests showed that these settings provide a good soil disturbance.

A genetic-designed beta basis function neural network for multi-variable functions approximation

We propose two evolutionary neural network-training algorithms for Beta basis function neural networks (BBFNN). Classic training algorithms for neural networks start with a predetermined network structure, and so the quality of the response of the BBFNN depends strongly on its structure. Generally, the network resulting from learning applied to a predetermined architecture is either insufficient or over- complicated. This paper describes two genetic learning models of the BBFNN. The first continuous genetic model changes the number of neurons in the hidden layer through the application of specific genetic operators. Each network is coded as a variable length string and some new genetic operators are proposed to evolve a population of individuals. A function is proposed to evaluate the fitness of individual networks. Applications to function approximation problems are considered to demonstrate the performance of the BBFNN and of the evolutionary algorithm. For the second discrete genetic model, each network is coded as a matrix for which number of rows is equal to the number of parameters in the function that will be approximated. The genetic algorithm operators change the number of neurons in the hidden layer. Some applications to functions with one and two parameters are considered to demonstrate the performance of the genetic model and the ability of genetic algorithm to be used for the design of BBFNN.

Numerical analysis solving the elastic dynamic problem

The purpose of this work is to present a numerical method for solving a problem of dynamic structures. For that, we use the Newmark algorithm and the finite element method for time and space discretisations, respectively. We prove the efficiency of the Newmark scheme in which the sum of elastic and kinetic energies is conserved. We conduct also numerical computations aiming to determine the influence of numerical parameters in terms of time step and mesh size on energy conservation.

SIMULATION OF MANUFACTURING CELLS WITH UNRELIABLE MACHINES

Cellular manufacturing is an application of a group technology used to improve the performance of manufacturing systems. A number of factors, including vulnerability to machine breakdown, under utilization of resources and eventual unbalanced workload distribution in a multi-cell plan disturb the smooth working of the factory when using the group technology concept. This paper focuses on a manufacturing cell composed of unreliable machines. We are interested in the problem of cell production availability facing unexpected circumstances due to an internal perturbation caused by machine breakdown. We consider a policy of intercellular transfer in the event of breakdown to improve the availability of the cells. We examine through simulation the performance of the system and evaluate the intercellular transfer policy in terms of some selected criteria. The results indicate, under the assumed conditions, that the developed policy improves the performance of the production cells. (Received in November 2005, accepted in August 2006. This paper was with the authors 2 months for 1 revision.)

Dynamic balance of a bipedal robot with trunk and arms subjected to 3D external disturbances

This paper deals with a novel approach based on the synergy between the dynamic motions of balancing masses and arms to reject large perturbations applied to the upper part of a bipedal robot. Initially the stabilization is carried out with a trunk having 4 degrees of freedom: three translations and one rotation. In the second time the stabilization is performed with a system with arms and having 10 DOFs. At first, for a vertical posture of the robot, the trunk bodies of ROBIAN are used to compensate the external three-dimensional efforts applied to the robot. During the simulation, this study allows us to determine on-line, the required movements and accelerations of the trunk bodies in order to maintain the robot stability. In the following stage, we study the required movements of a system which is made up of a trunk and two arms in order to ensure the robot stability in presence of disturbances or during a handling of an object. The same formalism is chosen to study the dynamics of the new upper part of the robot. This study shows the importance of the arms for the robot stability.

Evolutionary approach for the beta function based fuzzy systems

We propose an evolutionary method for the design of Beta fuzzy systems (BFS). Classical training algorithms start with a predetermined number of fuzzy rules for fuzzy systems. Generally speaking, the fuzzy system created is either insufficient or over-complicated. This paper describes a hierarchical genetic learning model of the BFS. In order to examine the performance of the proposed algorithm, it is used for the identification of an induction machine fuzzy plant model. The results obtained have been encouraging.

Quasi Static Fracture – Global Minimizer of the Regularized Energy

Fracture mechanics has been revisited aimed at modeling brittle fracture based on Griffith viewpoint. The purpose of this work is to present a numerical computational method for solving the quasi static crack propagation based on the variational theory. It requires no prior knowledge of the crack path or of its topology. Moreover, it is capable of modeling crack initiation. At the numerical level, we use a standard linear (P1) Lagrange finite element method for space discretization. We perform numerical simulations of a piece of brittle material without initial crack. We show also the necessity of adding the backtracking algorithm to alternate minimizations algorithm to ensure the convergence of the alternate minimizations algorithm to a global minimizer.