In the gradient based learning algorithms, the momentum has usually an improving effect in convergence rate and decreasing the zigzagging phenomena but sometimes it causes the convergence rate to decrease. The Parallel tangent (Partan) gradient is used as deflecting method to improve the convergence. In this paper, we modify the gradient Partan algorithm for learning the neural networks by using two different learning rates, one for gradient search and the other for accelerating through parallel tangent, respectively. Moreover, the dynamic self adaptation of learning rate is used to improve the performance. In dynamic self adaptation, each learning rate is adapted locally to the cost function landscape and the previous learning rate. Finally we test the proposed algorithm, called accelerated Partan on various problems such as xor and encoders. We compare the results with those of the dynamic self adaptation of learning rate and momentum.
Throughout the history of power system protection, improvements have continuously been made to most aspects of the protection system. However the choice of the protection function and the type selection are still carried out by specialist based on their personal experience. Since the applied design methods are strongly impacted by the personal philosophy of the designer, most of the existing protection system are not always optimal. In this paper a new method will be presented which is based on exact rules of the graph theory to achieve (at least near to) an optimum protection system for an arbitrary network configuration.
Up to now the design of power system protection is done by experts. Despite using software tools like relay coordination program, the design process is mostly based on personal experience. Therefore the design of the most existing protection systems are not always optimal. In this paper a new method will be presented which is based on exact rules for obtaining an optimal design.
IEEE Power Engineering Review (02721724)18(6)pp. 43-45
The coordination of protective relays of multiloop networks is a tedious and time consuming process. The complicated part of this process is the determination of a set of relays referred to as the break-point set (BPS), with a minimal size to start the coordination procedure. This letter introduces a new graph-theoretical approach to determine the BPS. This method reduces the complexity of the problem, by exploiting the available sparsity of the dependencies among relays in protection systems.
EUT Report, Eindhoven University of Technology, Faculty of Electrical Engineering (09298525)(98 -E-309)
The basic task of a electrical power system protection is to detect faults in system components, and to rapidly isolate them by opening all incoming current paths. On the other hand, in the case of any fault, the protection system should disconnect only a minimum number of components, to minimize interruptions to consumers. Therefore, relays of a protection system should be adjusted for the minimum possible operation time, while maintaining coordination among all relays. Modern power system networks are often multi-loop structured. Coordinated setting of protective relays in such networks is tedious and time consuming. The complicated part of this problem is the determination of a minimum set of relays, the so-called minimum Break Point Set (BPS), to start the coordination procedure. This report presents a new graph-theoretical method to determine a minimum or a near-to-minimum BPS. Using the rules of this method, the determination of a minimum BPS can be reduced and decomposed into sub-problems. Owing to the efficiency of these rules, the presented method quickly achieves the result, even for large networks. Moreover, due to the simplicity of the method, it can be manually applied to the graph of each network. Since the presented rules are general, they can be applied to improve any method dealing with BPS determination.
A methodology is proposed to simulate and calculate the electric field inside enclosures. Use of this method not only gives information over a wide frequency range and for various apertures, but also predicts the variation in the electric field inside the enclosure. The method is less time-consuming than other methods.
Modern power system networks often consist of multi-loop structures. The coordination of protective relays in such networks is an iterative and time consuming process. To minimize the number of iterations, a proper set of relays, referred to as Break Point Set (BPS), with a minimal size is required to start the coordination procedure. This paper presents a new graph-theoretical method for the BPS determination. The method represents the primary/backup relation among relays by a directed graph, referred to as dependency-diagram. This representation converts the BPS determination to a problem of graph theory. The method approaches step by step to a minimum and/or a near-to-minimum BPS, choosing the best break point relay at each step. The graph-theoretical rules of the method exploit the sparsity of the relations among the relays and reduce the complexity of the problem. Due to the generality of the method, it can be easily applied to any protection schemes and network configurations.
International Journal of Applied Electromagnetics and Mechanics (13835416)12(3-4)pp. 211-217
This paper addresses the modular decoupling methodology (MDM) [1]. It predicts EMC by the simulation of the effects of radiated emissions on an enclosure with a prescribed aperture and some contents. The proposed method could be a tool for preventing some possible EMI problems at the design stage. Computer simulations are developed using the finite element method and the investigation assesses the effectiveness of the shielding and susceptibility predictions. Exploration on an external emission source and its impact to lower stage is also considered.
In this paper, we propose to use a practical low rate error-correcting code in a time-hopping spread spectrum multiple access system employing a binary PPM signaling. To this end, superorthogonal codes, which are near optimal and have relatively low implementation complexity, are considered. The performance analysis shows that the proposed coded scheme outperforms the uncoded scheme significantly, or more importantly, at a given bit error rate the proposed scheme increases the number of users by a factor which is logarithmic in the number of pulses used in time hopping spread spectrum system.
IEE Proceedings: Vision, Image and Signal Processing (1350245X)147(3)pp. 231-237
It is shown that two algorithms obtained by simplifying a Kalman filter considered for a second-order Markov model are H∞ suboptimal. Similar to least mean squares (LMS) and normalized LMS (NLMS) algorithms, these second order algorithms can be thought of as approximate solutions to stochastic or deterministic least squares minimization. It is proved that second-order LMS and NLMS are exact solutions causing the maximum energy gain from the disturbances to the predicted and filtered errors to be less than one, respectively. These algorithms are implemented in two steps. Operation of the first step is like conventional LMS/NLMS algorithms and the second step consists of the estimation of the weight increment vector and prediction of weights for the next iteration. This step applies simple smoothing on the increment of the estimated weights to estimate the speed of the weights. Also they are cost-effective, robust and attractive for improving the tracking performance of smoothly time-varying models.
Moallem, M., Mirzaeian, B., Mohammed O.A., Lucas, C.
Publication Date: 2001
IEEE Transactions on Magnetics (00189464)(5 I)
In this paper, a novel multi-objective optimization method based on Genetic-Fuzzy Algorithm (GFA) is proposed. GFA is applied to optimize the five PI controller gains in the Indirect Field Oriented Control (IFOC) of an induction motor drive. The PI controller gains are designed to optimize the step response of the system. Rise-time, maximum over-shoot, settling time and steady state error are the objective functions. In this drive system, the simultaneous estimation of the rotor speed and time constant for a voltage source inverter-fed induction motor is discussed. The theory is based on the parallel Model Reference Adaptive Control System (MRAC). The vector control of the induction motor may be achieved in the rotor-flux-oriented frame. Furthermore, to eliminate the offset error caused by the change in the stator resistance, a fuzzy resistance is also designed. The simulation results of the new method for induction motor speed control is compared with the results obtained by the conventional method, which shows better performance.
This paper introduced an experimental evolution of the effectiveness of utilizing various moments as pattern features in human face technology. In this paper, we apply Pseudo Zernike Moments (PZM) for recognition human faces in two-dimensional images, and we compare their performance with other type of moments. The moments that we have used are Zernike Moments (ZM), Pseudo Zernike Moments (PZM) and Legendre Moments (LM). We have used shape information for human face localization, also we have used Radial Basis Function (RBF) neural network as classifier for this application. The performance of classification is dependent on the moment order as well as the type of moment invariant, but the classification error rate was below %10 in all cases. Simulation results on face database of Olivetti Research Laboratory (ORL) indicate that high order degree of Pseudo Zernike Moments contain very useful information about face recognition process, while low order degree contain information about face expression. The PZM of order of 6 to 8 with %1.3 error rate are very good features for human face recognition that we have proposed.
Conference Record - International Conference on Communications (05361486)10pp. 3017-3021
In [1], an ultra-wide bandwidth time-hopping spread-spectrum code division multiple access system employing a binary PPM signaling has been introduced, and its performance was obtained based on a Gaussian distribution assumption for the multiple access interference. A coded scheme of this system has been proposed in [2], which shows much better performance in comparison to the uncoded scheme under the above Gaussian assumption. In this paper, we present a more precise performance analysis of the system for both coded and uncoded schemes. Our analysis shows that the Gaussian assumption is not accurate for predicting bit error rates at high data transmission rates.
