نشریه مهندسی برق
سال چهل و یکم شماره 1 (پیاپی 61، بهار و تابستان 1390)

In this paper an environment implemented for performance modeling of on chip communication architecture is presented. Our goal in this development is providing an environment for fast on chip communication architecture modeling and simulation. In this environment the computation details of the components are ignored and they are modeled with a general timing parameter in a higher level of abstraction. The implemented environment generates an abstract dynamic model of communication architecture in OSCI TLM-2.0 and simulates it using SystemC simulation kernel.The most important advantage of this environment is its speed in simulation of communication architectures. On one hand, the computation details are ignored and on the other hand, the model is built and simulated in C++ without any other software interfaces

In this paper the effect of the distributed phase-shifted grating on the output power and longitudinal internal optical field of erbium doped DFB (Distributed Feedback) fiber lasers has been analyzed theoretically. It is found that the distributed phase shift along the grating significantly reduces the nonuniformity of longitudinal internal optical field and improves the probability of spatial hole burning (SHB) in DFB fiber lasers. DFB fiber lasers with distributed phase shift grating have higher threshold pumping and lower output power in comparison with a single -phase shift structure. On the other hand, numerical analysis shows that the structures with non-zero reflectance ends have higher output power. So in this work, the output power and flatness of internal optical field of DFB fiber lasers have been optimized by the reflectivity’s of two fiber ends and distributed phase-shifted grating structure. The theoretical model is based on the self-consistent solution of coupled mode equations and erbium ions concentration rate equations by transfer matrix method.

Knowledge of the styles of writing letters in the word is necessary in recognition of online (or offline) handwriting. In this paper, we have chosen Tarbiat Modarres University (TMU) dataset which includes about 10,000 samples from the 1000 Farsi useful subwords. Different grouping based on the main body and the small signs of the subwords have been introduced. Diverse styles of Farsi handwriting were investigated and the most important forms of letters writing were found. Then, the subwords were classified according to their styles. Frequency of letters that are written in different forms was discussed and analyzed. The results showed effective improvement in recognition rate using grouping based on the writing styles of letters.

The nonlinear portion of the electrical loads has increased significantly in recent years. Harmonic currents injected by these loads into the distribution systems should be considered when solving the capacitor placement problem in order to assure that the optimal solution does not result in excessive harmonic distortion or resonant conditions. This paper presents a binary-PSO approach for placement of the capacitors in a harmonics-polluted distribution system. The binary- PSO is integrated with a harmonic power flow algorithm. In this problem objective function is the total cost of the system power losses, cost of the capacitors placement and additional penalty terms.Constraints in this optimization procedure are the rms voltage and the voltage total harmonic distortion (THDv), as well as available capacitor units to be placed at each bus. Sensitivity analysis is used to find the candidate buses for capacitor locations. Proposed methods are tested on the 9 and 13 buses harmonic distribution systems. Numerical results show that the proposed method advantageous results for the capacitor placement problem.

The “Gaze Estimation” problem, due to its manifold applications, including human computer interaction for the handicapped, has been a topic of research since many years ago. Recently, the non-intrusive methods, based on image processing, are interested. In this paper, iris boundaries extraction, ellipse fitting to iris edges and geometrical relation between 3D and 2D images are used to estimate gaze direction. The accuracy of this method mainly relies on the accuracy of the fitted ellipse parameters which in turn requires higher resolution of iris image and/or more accurate algorithms. The key contributions of this paper are proposing novel algorithms to improve the performance of iris boundary extraction and ellipse fitting, while a conventional camera is employed. The main characteristics of the proposed algorithm for iris boundary extraction are being faster and robust to noisy and imperfect images. The ellipse fitting algorithm based on the proposed constraints, utilizes most of the iris boundary points which results in higher accuracy for the overall system. Moreover, this algorithm, unlike previous ones, has no pre-assumption about the head pose. Thus, another important contribution of this paper is the robustness of the proposed system to the head pose variations. The performance of the system has been evaluated on both synthetic and real images.

In this paper, an intelligent neural network method for controlling of a variable mass underwater vehicle is presented. The control scheme is capable of learning and adapting to changes in the vehicle dynamics and parameters, and consists of a gain tuning neural network and a variable gain PID controller. This neural network is trained so that the error between the plant output and reference signal is minimized. For updating the neural network weights, the system jacobian must be known. Because of the complexity and unknown governing equations of the system, jacobian calculation is too difficult and sometimes impossible. To overcome this problem another neural network (NNM) was trained for modeling the system and estimating the system jacobian. The results of this control scheme are compared with a constant gain PID controller. It is shown that the presented control scheme is better and more robust against disturbance than the conventional controller.

Nowadays, many techniques based on the heuristic optimization algorithms have been presented for multi-objective optimization problems. Although these techniques don’t guarantee to find the Pareto optimal front, they quest to achieve a good approximation of this set. In this paper, a technique based on gravitational search algorithm (GSA) is presented for optimization of multi-objective problem named multi-objective optimization gravitational search with non-dominated fronts (NFMOGSA). In the proposed method, the non-dominated front and the crowding distance of solutions are used to fitness assignment and diversity preservation. To evaluate the proposed algorithm, some experiments are done and the NFMOGSA is applied to the standard benchmark functions, SCH, KUR, FON, POL, ZDT1, ZDT2, ZDT3 and ZDT6. The results of applying NFMOGSA on the standard benchmark functions and comparison with other well known multi-objective optimization techniques confirm the efficiency and effectiveness of our technique in the multi-objective optimization problems.