IRAQI JOURNAL OF COMPUTERS, COMMUNICATIONS, CONTROL AND SYSTEMS ENGINEERING

This paper introduces the Slice Genetics Algorithm SGA which represents the proposed modification to the classic Genetic Algorithm GA scheme. The proposed algorithm has reduced the population size and maximum iteration in order to get fast and an optimal solution. This algorithm has been used for determining the optimal proportional- integral- derivative PID controller parameters. The proposed algorithm has versatile features, including, fast, stable rate convergence characteristic also it has good computational efficiency in improving the dynamic behavior for the system in term of reducing the maximum overshoot, rise time, settling time and steady-states error. The algorithm not only has benefit to improve the convergence characteristic, accuracy but it also shortened the processing time towards the optimal value based reducing the number of iteration from 40 to 4 or 6 iteration as clear in the MATLAB simulation results..

The technique of fusing a panchromatic (Pan) SAR image that has a high- spatial and low-spectral resolution with multispectral (MS) SAR images that have a low-spatial and high spectral resolution is very useful in many remote sensing applications that require both high-spatial and high-spectral resolution. In this paper, method for fusion SAR image is proposed based on framelet transform and new selection rule. The framelet transform is nearly shift-invariant with desired properties, short support, and symmetry. In the selection rule of proposed method, max rule is replaced with new relation depending on input SAR image. The proposed method is compared with other method such as HIS, PCA and wavelet methods. A quality of fused image is calculated based on the combination entropy, the correlation coefficient and the peak signal to noise ratio. It is showed from simulation result the quality measured for proposed method can indicate the information content of the fused image is higher compared to the information content of the input panchromatic and multispectral images, also its noticed the proposed method provides richer information comparing with other methods.

In this paper, direct neural controller for braking system is proposed. Learning of the presented controller depends on the training data that comes from running the switching gain controller at different conditions of drive. The training data consist of relative velocity error, distance error and braking force. The feed-forward neural network is used to build direct neural controller with two hidden layers and using back-propagation training algorithm. The performance of the presented controller is validated using nonlinear braking model. Simulation results show the presented controller is able to prevent the collision of vehicles at different driving conditions. Also, the results show superiority of the direct neural controller in comparison with the switching gain controller at all drive cases that are tested in this work.

The ability to recognize quickly and accurately which we encounter is fundamental to normal intelligent human behavior. However, how the learning of categories which objects in the world fit into takes place is still an unanswered question. One thing is certain though; much of the learning that takes place allows humans to cope with the changing they encounter. One of the most important aspects of human intelligence is its flexibility which has allowed humans to prosper in a dynamic world. Humans do not suffer from the ills of old fashioned hard rule based artificial intelligence. The study tested six cubes. The vertices of the cubes represent individual stimuli constructed from three binary dimensions. The dimension of the stimuli can be assumed to correspond to shape (square vs. circle), color (black vs. white), and size (large vs. small). Four stimuli belonged to one category and the other four to a different category. These constraints result in six problem types, which are illustrated by the six cubes. The circle vertices represent stimuli that belong to category A, and the square vertices represent stimuli that belong to category B. The faces of the cubes represent a constant value across one of the three dimensions that define the stimuli. This work presents experiments with two different classifier systems: learning when fitness is based upon strength and specificity, and learning when fitness is based on strength alone. The system is implemented using Pascal programming language. Results show lower performance of the system when depending on strength alone. By contrast, the run with strength and specificity allows a fast desired output.

Depending on the response of the system, digital Filters can be designed using frequency sampling or windowing methods; but these methods have a problem in precise control of the critical frequencies. In the sampling method, the weighted approximation error between the actual frequency response and the desired filter response is spread across the pass-band and the stop-band and the maximum error is minimized, resulting ripples in the pass-band and the stop-band. The frequency sampling method has the same tolerance requirements as the windowing method. In this work we implemented a digital FIR high pass filter using MATLAB program (FDATools) using sampling and windowing methods, then the design in the FPGA kit is downloaded by generating VHDL description. A comparison the amount of the component has been used in the FPGA for both methods. The FIR filter is implemented using Spartan 3AN- XC3S700a-4FG484 FPGA and simulated with the help of Xilinx ISE (Integrated Software Environment) Software WEBPACK Project Navigator 11i.

Artificial Neural Networks (ANN) can be used as intelligent controllers to control non-linear dynamic systems through learning, which can easily accommodate the non linearity's, time dependencies, model uncertainty and external disturbances. Modern power systems are complex and non-linear and their operating conditions can vary over a wide range. The Nonlinear Auto-Regressive Moving Average (NARMA-L2) model system is proposed as an effective neural networks controller model to achieve the desired robust Automatic Voltage Regulator (AVR) for Synchronous Generator (SG) to maintain constant terminal voltage. The concerned neural networks controller for AVR is examined on different models of SG and loads. The results shows that the neuro-controllers have excellent responses for all SG models and loads in view point of transient response and system stability compared with conventional PID controllers. Also shows that the margins of robustness for neuro-controller are greater than PID controller.

Pattern recognition problems computer based are very important and essential in our real life. There are many approaches have been used in pattern recognition problem such as: Fourier Descriptor, Moment Invariant. But the main defect of these methods is the long time processing and large computer space. This paper, presents a new approach Artificial Intelligence, of Rule Induction technique. By this approach, the essential and specific features of object have been extracted from contour of object to be recognized. The characteristic of these features are easy computed and requires fewer amounts of time and space, then high speed in recognition and decision. Such features are (number of curves inside the fingerprint, number of check point for each curve). It gives good and accurate results. We test the performance of this system using many contours of fingerprint, and get good and accurate results.