Joseph A. Jervase

Solar radiation estimation using artificial neural networks

Artificial Neural Network Methods are discussed for estimating solar radiation by first estimating the clearness index. Radial Basis Functions, RBF, and Multilayer Perceptron, MLP, models have been investigated using long-term data from eight stations in Oman. It is shown that both the RBF and MLP models performed well based on the root-mean-square error between the observed and estimated solar radiations. However, the RBF models are preferred since they require less computing power. The RBF model, obtained by training with data from the meteorological stations at Masirah, Salalah, Seeb, Sur, Fahud and Sohar, and testing with those from Buraimi and Marmul, was the best. This model can be used to estimate the solar radiation at any location in Oman.

Solar cell parameter extraction using genetic algorithms

In this paper, a technique based on genetic algorithms is proposed for improving the accuracy of solar cell parameters extracted using conventional techniques. The approach is based on formulating the parameter extraction as a search and optimization problem. Current–voltage data used were generated by simulating a two-diode solar cell model of specified parameters. The genetic algorithm search range that simulates the error in the extracted parameters was varied from ± 5t o±100% of the specified parameter values. Results obtained show that for a simulated error of ±5% in the solar cell model values, the deviation of the extracted parameters varied from 0.1 to 1% of the specified values. Even with a simulated error of as high as ±100%, the resulting deviation only varied from 2 to 36%. The performance of this technique is also shown to surpass the quasi-Newton method, a calculus-based search and optimization algorithm.

Monthly average daily solar radiation and clearness index contour maps over Oman

Using meteorological data spanning more than ten years in Oman, radial basis function neural networks were developed for estimation of the solar radiation and clearness index for any location in Oman, given its longitude, latitude, altitude, sunshine hours and month of the year. Data generated for 25 different locations were then used to plot contour maps for monthly average daily solar radiation and clearness index over Oman. These maps will serve as a much needed reference for design and performance evaluation of solar energy conversion systems in this region.

Characteristic analysis of resonant-cavity-enhanced (RCE) photodetectors

Analyses of light field distribution within a lossy resonant-cavity-enhanced (RCE) p-i-n photodetector is performed. A new expression for the field distribution inside the cavity is derived. This expression is used to derive an expression for the quantum efficiency taking into account the standing wave effect. The field distribution expression is also used in the continuity equations to derive and evaluate the frequency response of the RCE photodetector.

Contour maps for sunshine ratio for Oman using radial basis function generated data

Contour maps for sunshine hours and sunshine ratios for Oman have been generated. The data to generate these maps were obtained using an RBF neural network model. This model estimates sunshine hours and ratios for a given point based on its latitude, longitude, altitude and month of the year. Data from 25 locations were used to plot the contour maps. These maps provide a needed reference for the spatial distribution of sunshine hours and sunshine ratios on a monthly basis for the whole of Oman from which estimates can be made for any location.

Classification of modulation signals using statistical signal characterization and artificial neural networks

Modulation recognition systems have to be able to correctly classify the incoming signal modulation scheme in the presence of noise. A new method for classification of analogue and digital modulated signals at low signal-to-noise ratio (SNR) is introduced in this paper. This method uses the statistical signal characterization (SSC) to extract parameters to classify the different modulation signals. The SSC technique produces a set of four numerical parameters for a specific modulated signal. Subsequent comparison of these parameters to those of other waveforms provides the basis for our classification system. The results of SSC technique are applied to an artificial neural network (ANN) to have a robust classification system in the presence of noise down to SNR of 3dB. No a priori information is required by this technique about the set of input waveforms. The input to the classification system can be analogue or digital signals or a combination of both. The proposed technique shows a 100% efficiency of classification of analogue signals or digital signals at SNR of 7dB. This classification efficiency reduces to 83% and 86% for analogue or digital signals at SNR of 3dB. The SSC technique shows better classification results in comparison with other techniques with an important advantage over other methods, which is the simplicity of the neural network needed with this technique due to the small number of features used in the classification.

Design of resonant-cavity-enhanced photodetectors using genetic algorithms

Expressions of quantum efficiency of resonant-cavity-enhanced (RCE) PIN photodetectors reported in the literature are based on the assumption of constant reflectivities of the quarter-wave stacks (QWS) at the ends of the cavity. The quantum efficiency is formulated in a closed analytical form that includes the structural parameters of the photodetector and takes into account the wavelength dependence of the reflectivities and the active region absorption coefficient. The variation of the QWS reflectivity and, in particular, its phase constant with wavelength has a significant influence on the resulting quantum efficiency spectra, as demonstrated in this paper. The results are in very good agreement with recently published experimental data which show a dominant peak at the operating wavelength. This behavior has not been predicted by previous simulation results. Since the quantum efficiency spectra are not periodic, the use of the finesse, defined as the ratio of the free spectral range to the full width at half maximum, as a measure of wavelength selectivity is not valid. The conventional quality factor definition used for filter design is thus adopted as a measure of selectivity. A genetic algorithm-based optimization and design procedure for RCE photodetectors have also been developed with the quantum efficiency, quality factor, and frequency bandwidth as input design parameters.

Design of ultra-fast dual-wavelength resonant-cavity-enhanced Schottky photodetectors

A systematic optimization procedure for the design of RCE Schottky photodetectors to achieve maximum quantum efficiency and high speed operation at 1.3 and 1.55 /spl mu/m wavelengths is presented. The quantum efficiency formulation used includes the structural parameters of the photodetector and takes into account the wavelength dependence of the top and bottom mirrors reflectivities. The results have shown that the value of the thickness of the antireflection coating layer has a major influence in selecting the width of the photodetector to simultaneously achieve maximum quantum efficiency and high bandwidth at the two wavelengths. Simulated values of 270 and 40 GHz were obtained, respectively, for the 3-dB carrier-transit time-limited bandwidth and bandwidth-efficiency product for an RCE Schottky photodetector with a 0.02-/spl mu/m gold layer.

A micro-genetic algorithm-based CDMA multi-user detector

A new multi-user detection technique that utilizes only the micro-genetic algorithm (/spl mu/GA) is developed. In contrast to previous approaches, the micro-genetic algorithm is not hybridized with any detector other than a conventional detector (CD) or a decorrelator as an initial stage. The use of this algorithm considerably reduces the number of operations as well as the total cost of the proposed detector. A 5-user DS-CDMA system is adopted to examine the performance of the /spl mu/GA detector. Using computer simulations, it is shown that the proposed /spl mu/GA receiver achieves a performance close to the optimal detector as measured by the bit-error-rate (BER). This performance is achieved at a much lower computational time compared to the optimal detector. The potential of the /spl mu/GA detector as a suboptimal DS-CDMA multi-user detector is thus demonstrated.

Optimization procedure for the design of ultrafast, highly efficient and selective resonant cavity enhanced Schottky photodiodes

An expression of quantum efficiency for high-speed resonant-cavity-enhanced (RCE) Schottky photodiodes is derived. This expression includes the structural and the physical parameters of the photodetector and takes into account the parameters of the metallic Schottky mirror and the wavelength dependence of the reflectivities. The metal layer thickness sets the maximum achievable quantum efficiency as it decays exponentially with it. The antireflection coating layer, on the other hand, determines the photodetector selectivity and the optimum absorption layer thickness that maximizes its quantum efficiency. An algorithm for the design and optimization of RCE Schottky photodetectors has been developed. Theoretical values of 647 GHz and 129 GHz were obtained, respectively, for the carrier-transit time limited 3-dB bandwidth and bandwidth-efficiency product for an RCE Schottky photodetector with a 0.02 /spl mu/m gold layer.