Showing papers by "Mahdi Aliyari Shoorehdeli published in 2014"

Implementation and Control of X–Y Pedestal Using Dual-Drive Technique and Feedback Error Learning for LEO Satellite Tracking

Abstract: This brief presents a X–Y pedestal using the feedback error learning (FEL) controller with adaptive neural network for low earth orbit (LEO) satellite tracking applications. The aim of the FEL is to derive the inverse dynamic model of the X–Y pedestal. In this brief, the kinematics of X–Y pedestal is obtained. To minimize or eliminate the backlash between gears, an antibacklash gearing system with dual-drive technique is used. The X–Y pedestal is implemented and the experimental results are obtained. They verify the obtained kinematics of the X–Y pedestal, its ability to minimize backlash, and the reduction of the tracking error for LEO satellite tracking in the typical NOAA19 weather satellite. Finally, the experimental results are plotted.

Feature subset selection and parameters optimization for support vector machine in breast cancer diagnosis

Abstract: Due to high death rate in women with breast cancer, the detection will play a major role in the treatment of this type of cancer. Therefore, the early detection of breast cancer will increase the patients' chances of survival. The main tendency in feature extraction has been illustrating the data in a lower dimensional and different feature space, for instance, using principal component analysis (PCA). In this paper, we argue that feature selection depend on top of eigenvalue certainly is not proper because they may not encode useful information for classilcation purposes, features should be selected form all the components by feature selection methods. So, Genetic Algorithm (GA) is used in the most favorable selection of principal components instead of using classical method. We have applied PCA for dimension reduction, genetic algorithms for feature selection and support vector machines for classification. The estimate of this Algorithm has been done based on Wisconsin Breast Cancer Dataset (WBCD) which is commonly used among researchers who use machine learning methods for breast cancer diagnosis. The performance of this approach is given. In addition, the methods used in the past have been compared to the performance of the chosen approach. This approach affords optimal classification which is capable to minimize amount of features and maximize the accuracy sensitivity, specificity and receiver operating characteristic (ROC) curves. 10-fold cross-validation has been used on the classification phase. The average classification accuracy of the developed PCA+GA+SVM system is obtained 100% for a subset that contained two features. This is very favorable compared to the previously reported results.

Prediction of seasonal influenza epidemics in Tehran using artificial neural networks

Abstract: Prediction of seasonal influenza epidemics is certainly a forming and effective step towards taking appropriate preventive actions. Improvement on public informing, decreasing the number of infected cases, undesirable effects and deaths due to influenza and also increasing vigilance of Iranian Influenza Surveillance System (IISS), have been practical goals of this research. A forecasting system has been designed and developed using Artificial Neural Networks (ANNs). It is a novel research as a novel dataset has been exploited. The data are categorized in two groups of climatic parameters (temperature, humidity, precipitation, wind speed & sea level pressure) and number of patients (number of total referrals and number of patients with Influenza-Like Illnesses (ILI)). In order to evaluate the model performance, different cost functions are defined and results indicate that the model provides the possibility of a satisfactory forecasting and is practically helpful to achieve the objectives already claimed.

Nonlinear system identification of Hammerstein-Wiener model using AWPSO

Abstract: This paper presents the problem of constructing an appropriate model with Hammerstein-Wiener structure for nonlinear system identification. In this structure, the nonlinearity is implemented through two static nonlinear blocks where a linear dynamic block is surrounded by two nonlinear static systems. Algorithms such as genetic algorithm can find unknown parameters, but the complexity of the calculations is their weakness. Hence, a class of computational methods named Particle Swarm Optimization (PSO) is used. To avoid trapping in local optimum and improve performance; Adaptive Weighted Particle Swarm Optimization (AWPSO) method is used. The training method is responsible for finding the optimal values of the parameters of the transfer function from the linear dynamic part as well as the coefficients of the nonlinear static functions.

Real-time dynamic HIL simulator of gas turbine, governor, generator and grid for static excitation of a 200-MVA synchronous generator

Abstract: A real-time dynamic hardware-in-loop (HIL) simulator of an RTX real-time subsystem (RTSS) was developed by using LabVIEW (G language). The main idea of this work was to determine the feasibility and accuracy of widely available and highly competitive commercial products, such as personal computers on an RTSS, as an alternative to conventional prohibitive real-time simulators in dynamic studies of power systems. The implemented system is a self-contained heavy-duty gas turbine, governor, synchronous 200-MVA, 15.75-kV machine and a simplified electrical network. The HIL simulator was customized to interact with a 1518-kW static exciter. The role of this HIL simulator is to provide real-time digital and analog signals for static exciter systems (SES) and to simulate the mechanical and electrical components in a closed-loop, fixed-step solver applied by a well-known numerical solution method. This sophisticated yet exceptionally economic HIL simulator provides engineers with a safe environment to analyze the dynamic performance of static exciters and investigate their natural restraints and functionalities. It also provides a safe environment to analyze some naturally destructive tests.

Neural Networks for Normative Knowledge Source of Cultural Algorithm

Abstract: This study presents the normative knowledge source for the belief space of cultural algorithm(CA) based on an adaptive Radial Basis Function Neural Network (RBFNN). The use of the RBFNN makes it possible to use the previous upper and lower bounds of the normative knowledge to update them and to extract a logical relationship between the previous parameters of the normative knowledge and their new values. The proposed algorithm(N3KCA) is similar to what the human brain does, i.e. to predict the new values of the bounds of normative knowledge based on the previous ones and some knowledge, which it has gained from the previous successive updates. Finally, the proposed cultural algorithm is evaluated on 10 unimodal and multimodal benchmark functions. The algorithm is compared with several other optimization algorithms including previous version of cultural algorithm. In order to have a fair comparison, the number of cost function evaluation is kept the same for all optimization algorithms. The obtaine...

Multidisciplinary modelling and position tracking control of an electro-hydrostatic actuator using a novel adaptive fuzzy-PID controller

Abstract: Electro-hydrostatic actuator (EHA) is a kind of hydraulic system in which fluid is routed directly by pump to the actuator. In this study, a novel adaptive fuzzy-PID controller is developed to improve position controlling performance of an EHA. First of all, design and simulation of an EHA by using multidisciplinary modelling method is presented. This model is evaluated by soft validation method. The whole proposed novel control system is composed of a pair of interconnected subsystems, that is, a simple fuzzy-PID controller (SFPID) and a radial basis function neural network (RBFNN) to enhance the tracking performance. The RBFNN fuzzy-PID control (RBFNNF-PID) is applied to EHA. Also, SFPID control, fuzzy-PID control based on extended Kalman filter using grey predictor (FPIDKG) and simple adaptive control (SAC) as significant controls are applied to EHA. The simulation results have shown a significant improvement in transient response and reduction in sum square error (SSE).

Evolutionary Algorithms to Compute the Optimal Parameters of Gaussian Radial Basis Adaptive Backstepping Control for Chaotic Systems

Abstract: In this paper, evolutionary algorithms are proposed to compute the optimal parameters of Gaussian Radial Basis Adaptive Backstepping Control (GRBABC) for chaotic systems. Generally, parameters are chosen arbitrarily, so in several cases this choice can be tedious. Also, stability cannot be achieved when the parameters are inappropriately chosen. The optimal design problems are to introduce optimization algorithms like Genetic Algorithms (GA), Particle Swarm Optimization (PSO) in order to find the optimal parameters which minimize a cost function defined as an error quadratic function. These methods are applied to two chaotic systems; Duffing Oscillator and Lu systems. Simulation results verify that our proposed algorithms can achieve enhanced tracking performance regarding similar methods.

Robust state fault diagnosis in nonlinear discretetime systems with modelling uncertainties; using an automated intelligent methodology

Abstract: Modern systems are required to guarantee a high degree of safety and self-diagnostics capabilities. This paper investigates the problem of state fault diagnosis in nonlinear systems with modeling uncertainties. In contrast with common literature, the fault diagnosis scheme is proposed in discrete time domain. This property relaxes the risk of stability and performance degradation in deriving discrete equivalent of continuous methods. An estimator is designed in order to generate residual signal by utilizing a proper nonlinear state transformation. A robust compensator term is implemented in estimator to decrease effect of modeling uncertainties and approximation error on residual signal. When the residual signal is exceeded detection threshold, an on-line fault approximator is turned on and trained by appropriate parameter update law. An extra term is considered in update rule to overcome the need of persistency of excitation (PE). The implement of all robust compensator term, PE relaxing term and proper parameter adaption law improve the accuracy of fault reconstruction. The result would be obviously vital in tolerant and time-life prediction stages after fault diagnosis.