Ahmad F. Al-Ajlouni

Approximation of nonlinear systems with radial basis function neural networks

A technique for approximating a continuous function of n variables with a radial basis function (RBF) neural network is presented. The method uses an n-dimensional raised-cosine type of RBF that is smooth, yet has compact support. The RBF network coefficients are low-order polynomial functions of the input. A simple computational procedure is presented which significantly reduces the network training and evaluation time. Storage space is also reduced by allowing for a nonuniform grid of points about which the RBFs are centered. The network output is shown to be continuous and have a continuous first derivative. When the network is used to approximate a nonlinear dynamic system, the resulting system is bounded-input bounded-output stable. For the special case of a linear system, the RBF network representation is exact on the domain over which it is defined, and it is optimal in terms of the number of distinct storage parameters required. Several examples are presented which illustrate the effectiveness of this technique.

ECG signal compression using combined modified discrete cosine and discrete wavelet transforms

A new hybrid two-stage electrocardiogram (ECG) signal compression method based on the modified discrete cosine transform (MDCT) and discrete wavelet transform (DWT) is proposed. The ECG signal is partitioned into blocks and the MDCT is applied to each block to decorrelate the spectral information. Then, the DWT is applied to the resulting MDCT coefficients. Removing spectral redundancy is achieved by compressing the subordinate components more than the dominant components. The resulting wavelet coefficients are then thresholded and compressed using energy packing and binary-significant map coding technique for storage space saving. Experiments on ECG records from the MIT-BIH database are performed with various combinations of MDCT and wavelet filters at different transformation levels, and quantization intervals. The decompressed signals are evaluated using percentage rms error (PRD) and zero-mean rms error (PRD1) measures. The results showed that the proposed method provides low bit-rate and high quality of the reconstructed signal. It offers average compression ratio (CR) of 21.5 and PRD of 5.89%, which would be suitable for most monitoring and diagnoses applications. Simulation results show that the proposed method compares favourably with various state-of-the-art ECG compressors.

A new developed educational approach to improve conventional teaching methodology of the power electronics laboratory

This paper proposes a new computerized educational approach to teach the power electronics laboratory. It describes PSpice implementation of the core power electronic circuits that depend on thyristor circuits to identify behaviors with load variations. It uses the developed simulation models to support and enhance power electronics education at the undergraduate level. These simulations successfully integrate the contents of the power electronic laboratory course. A study of the impact of these simulations on the results of the students showed that it helped them to master the course contents and to gain better grades.

Effect of instructional software on achievement of undergraduate students in Digital Signal Processing

Traditional tools used in education, are increasingly more complemented with computer based Instructional software. This paper presents an interactive approach for teaching the Digital Signal Processing (DSP) course. The effect of the Instructional Software on the achievement of students in the subject of DSP is compared to the conventional teaching method.

Identification of nonlinear discrete-time systems using raised-cosine radial basis function networks

An effective technique for identifying nonlinear discrete-time systems using raised-cosine radial basis function (RBF) networks is presented. Raised-cosine RBF networks are bounded-input bounded-output stable systems, and the network output is a continuously differentiable function of the past input and the past output. The evaluation speed of an n-dimensional raised-cosine RBF network is high because, at each discrete time, at most 2n RBF terms are nonzero and contribute to the output. As a consequence, raised-cosine RBF networks can be used to identify relatively high-order nonlinear discrete-time systems. Unlike the most commonly used RBFs, the raised-cosine RBF satisfies a constant interpolation property. This makes raised-cosine RBF highly suitable for identifying nonlinear systems that undergo saturation effects. In addition, for the important special case of a linear discrete-time system, a first-order raised-cosine RBF network is exact on the domain over which it is defined, and it is minimal in terms of the number of distinct parameters that must be stored. Several examples, including both physical systems and benchmark systems, are used to illustrate that raised-cosine RBF networks are highly effective in identifying nonlinear discrete-time systems.

Active control of narrowband acoustic noise of unknown frequency using a phase-locked loop

An adaptive feedforward technique for active control of narrowband acoustic noise of unknown frequency is presented. A phase-locked loop is used to determine the fundamental frequency of the periodic component of the primary noise. Adaptive control of the residual noise is achieved with a least mean square method. Real time implementation is facilitated with an efficient recursive technique that is used to update the controller gains at each time step. Significant noise cancellation is demonstrated in physical experiments, which include additive broadband noise.

Design of Immune Algorithm based two-dimensional recursive digital filters using multi-level Orthogonal Arrays

Taguchi Immune Algorithm (TIA) is based on both features of the biological immune system and the Taguchi method which increases the ability of the Immune Algorithm (IA) to find the global optimal solution in a nonlinear space. In the TIA, the clonal proliferation within hypermutation for several antibody diversifications and the recombination by using the Taguchi method for the local search are integrated to improve the capabilities of exploration and exploitation. Two major tools are used in the Taguchi method; namely the Orthogonal Arrays (OAs) and the Signal to Noise Ratio (SNR). The effect of selecting the number of levels adopted in the construction of OAs on TIA is not studied before. So, this paper addresses the problem increasing the convergence speed of immune algorithm based two-dimensional recursive digital filters design process by adopting two, three and four levels OAs. For seek of comparison, the same computational experiments adopted in [1] are considered. Numerical results show that increasing the number of OA levels yields to faster convergence and better antibody genes selection in order to achieve the potential recombination, and consequently enhance the design process.

Fuzzy Approach to Optimal Generation Scheduling for Gencos in Competitive Electricity Markets

Abstract This paper presents different fuzzy membership functions for maximizing the GenCo’s profit in competitive electricity markets. Rectangular, triangular and trapezoidal fuzzy membership functions are assigned for market prices and probability that reserves are called and generated. The proposed fuzzy membership functions produce an optimal profit fuzzy function that reflects the market uncertainties and helps in decision making. Transmission congestion is incorporated through locational marginal prices in the proposed formulation. IEEE 118-bus system is used to demonstrate the superiority of the proposed approach.

An ECG signal compressor based on the selection of optimal threshold levels of discrete wavelet transform coefficients.

Compression of electrocardiography (ECG) is necessary for efficient storage and transmission of the digitized ECG signals. Discrete wavelet transform (DWT) has recently emerged as a powerful technique for ECG signal compression due to its multi-resolution signal decomposition and locality properties. This paper presents an ECG compressor based on the selection of optimum threshold levels of DWT coefficients in different subbands that achieve maximum data volume reduction while preserving the significant signal morphology features upon reconstruction. First, the ECG is wavelet transformed into m subbands and the wavelet coefficients of each subband are thresholded using an optimal threshold level. Thresholding removes excessively small features and replaces them with zeroes. The threshold levels are defined for each signal so that the bit rate is minimized for a target distortion or, alternatively, the distortion is minimized for a target compression ratio. After thresholding, the resulting significant wavelet coefficients are coded using multi embedded zero tree (MEZW) coding technique. In order to assess the performance of the proposed compressor, records from the MIT-BIH Arrhythmia Database were compressed at different distortion levels, measured by the percentage rms difference (PRD), and compression ratios (CR). The method achieves good CR values with excellent reconstruction quality that compares favourably with various classical and state-of-the-art ECG compressors. Finally, it should be noted that the proposed method is flexible in controlling the quality of the reconstructed signals and the volume of the compressed signals by establishing a target PRD and a target CR a priori, respectively.

A Collocation-Based Algorithm for Analyzing Bifurcations in Phase Locked Loops with Tanlock and Sawtooth Phase Detectors

Analysis of bifurcation of second-order analog phase locked loop (PLL) with tanlock and sawtooth phase detectors is investigated. Both qualitative and quantitative analyses are carried out. Qualitatively, the basin boundaries of the attractors were constructed by plotting the stable and the unstable manifolds of the system. The basin boundaries show that the PLL under consideration for certain loop parameters has a separatrix cycle which terminates the limit cycle (out-of-lock state) and the loop pulls-in. This behavior is known in literature as homoclinic bifurcation and the value of the bifurcation parameter where this process occurs is called the pull-in range. Quantitatively, we propose a collocation-based algorithm to compute the separatrix cycle and the pull-in range. The separatrix cycle is approximated by a finite set of harmonics N with unknown amplitudes and by utilizing the fact that this limit cycle bifurcates from a separatrix cycle, a system of nonlinear algebraic equations is derived. For given values of filter parameters and gain, the algorithm numerically solves for the unknown amplitude of the harmonics and the value of the pull-in range simultaneously by evaluating the system at the collocation points. Results demonstrate that phase locked loop with sawtooth phase detector characteristics has the wider pull-in range followed by tanlock and sinusoidal, respectively.