Fast Convergent Algorithm Research Articles

Annual hourly CFD simulation: New approach—An efficient scheduling algorithm for fast iteration convergence

Computational fluid dynamics (CFD) is one of the branches of fluid mechanics that uses numerical methods and algorithms to solve and analyze problems that involve fluid flows. Annual hourly fast flow simulations are needed for some applications in building industry, such as the conceptual design of indoor environment, or coupled with energy simulation to provide deep analysis on the performance of the buildings. Year round simulation, which consists of 8760 (365×24) independent hourly simulations, is needed to help the designer investigate the problem clearly. However, CFD computation is time consuming, and usually only two or three extreme cases can be simulated in practice. Annual hourly simulation using the traditional method can be considered as a computational intractable problem. Based on previous researches, even though the fast fluid dynamics (FFD) algorithm combined with the General Purposed Graphics Processing Unit (GPGPU) hardware acceleration can make CFD simulation much faster (400x), annual hourly simulation still requires CFD performance to be further improved by 10-20x to make it practical. In this study, a minimal spanning tree based scheduling algorithm is developed, which always gives the best CFD simulation strategy that reuses previous calculated results to generate new results, thus making iteration convergence much faster. It is shown in the paper that the annual hourly simulation by GPGPU accelerated FFD by using this new algorithm requires a similar simulation time to the one used to perform two or three extreme cases of simulation using the traditional method.

The Numerical Simulation and Analysis of Wind Pressure Distribution on Sloping Roof of Storey Factory

As wind-induced damage always occurs at the roof of storey factory, the roof pressure of storey factory under different wind directions was analyzed based on the Realizable k-ε turbulence model. Hexahedral element based numerical modeling was chosen, and the pressure-linked equations were solved with fast-convergence SIMPLEC algorithm. The results showed the wind pressure at the leading edge from the windward side is far greater than other regions of the roof, indicating this area should be the main issue in wind-resistant design.

Stereo matching algorithm based on fast-converging belief propagation

Concerning the high computation complexity and low efficiency in traditional stereo matching method based on belief propagation,a fast-converging algorithm was proposed.When calculating the confidence level of each pixel,the algorithm only utilized the information translated from the neighboring pixels in an adaptive support window,while ignoring the impact of the pixels beyond the window.The experimental results show that the proposed algorithm can reduce 40% to 50% of computation time while maintaining the matching accuracy.Therefore,it can meet the real-time requirement for stereo matching.

Data-driven estimation of the inertia moment of wind turbines: A new ice-detection algorithm

Turbine blades accumulate ice under certain atmospheric conditions, such as low temperature and high humidity. This implies additional loads that might damage the turbine. A reliable ice-detection algorithm is required to shut down the turbine and prevent damages. A simple sensorless technique is proposed in this article for detection of both icing and ice shedding events. The technique is based on the estimation of the turbine inertia moment using generator speed measurements, turbine model and robust and fast convergent data-driven algorithms. Estimated inertia moment can be used for both ice detection and adaptation of the parameters of the control system. Implementation of this technique allows a stable turbine operation during hazardous ice conditions via adjustment of the control system parameters or turbine shut down in the extreme icing conditions.

A globally converging algorithm for adaptive manipulation and trajectory following for mobile robots with serial redundant arms

SUMMARYIn this paper the tip-over stability of mobile robots during manipulation with redundant arms is investigated in real-time. A new fast-converging algorithm, called the Circles Of INitialization (COIN), is proposed to calculate globally optimal postures of redundant serial manipulators. The algorithm is capable of trajectory following, redundancy resolution, and tip-over prevention for mobile robots during eccentric manipulation tasks. The proposed algorithm employs a priori training data generated from an exhaustive resolution of the arm's redundancy along a single direction in the manipulator's workspace. This data is shown to provide educated initial guess that enables COIN to swiftly converge to the global optimum for any other task in the workspace. Simulations demonstrate the capabilities of COIN, and further highlight its convergence speed relative to existing global search algorithms.

Synthesis of Split-Rings-Based Artificial Transmission Lines Through a New Two-Step, Fast Converging, and Robust Aggressive Space Mapping (ASM) Algorithm

This paper is focused on the synthesis of artificial transmission lines based on complementary split ring resonators (CSRRs). The considered structures are microstrip lines with CSRRs etched in the ground plane and microstrip lines loaded with both CSRRs and series capacitive gaps. An aggressive space mapping (ASM) optimization algorithm, able to automatically generate the layout of these artificial lines, has been developed. The tool has been optimized in order to achieve fast convergence and to provide accurate results. The main relevant aspects of the proposed algorithm (based on a novel two-step ASM optimization approach) are: 1) the capability to provide the implementable circuit elements of the equivalent circuit model of the considered artificial lines and 2) the ability to converge in a few (unprecedented) iteration steps, due to a new procedure to generate the initial layouts (which are very close to the final ones). First, the software is tested through the synthesis of several CSRR-based microstrip lines, and then some practical application examples of such artificial lines are reported to illustrate the potential of the proposed synthesis tool.

Revolute manipulator workspace optimization using a modified bacteria foraging algorithm: A comparative study

Robotic manipulators with three-revolute (3R) motions to attain desired positional configurations are very common in industrial robots. The capability of these robots depends largely on the workspace of the manipulator in addition to other parameters. In this study, an evolutionary optimization algorithm based on the foraging behaviour of the Escherichia coli bacteria present in the human intestine is utilized to optimize the workspace volume of a 3R manipulator. The new optimization method is modified from the original algorithm for faster convergence. This method is also useful for optimization problems in a highly constrained environment, such as robot workspace optimization. The new approach for workspace optimization of 3R manipulators is tested using three cases. The test results are compared with standard results available using other optimization algorithms, i.e. the differential evolution algorithm, the genetic algorithm and the particle swarm optimization algorithm. The present method is found to be superior to the other methods in terms of computational efficiency.

Modelling the nonlinear dynamic behaviour of a boiler-turbine system using a radial basis function neural network

SUMMARY Building an appropriate mathematical model that describes the system behaviour with a certain degree of satisfaction is quite challenging owing to the uncertain and volatile nature of thermodynamic constants and geometric parameters. In this paper, we present a technique to approximate and validate the dynamic behaviour of the Astrom–Bell boiler-turbine power plant based on an RBFNN over a large operating range. The proposed RBFNN is applied to solve the parametric identification problem for nonlinear and complex systems using an optimiser based on a hybrid genetic algorithm. This optimiser is composed of the gradient descent optimiser and a genetic algorithm for fast convergence. Two simulations were performed to show the effectiveness of the proposed technique under different situations with several boiler-turbine input variables. The optimal structure and parameters of the obtained RBFNN-based model emulates well the dynamic behaviour of the Astrom–Bell boiler-turbine system. Copyright © 2013 John Wiley & Sons, Ltd.

Fast Converging Generalized Turbo Decoding Scheme with Enhanced Throughput for Mobile Radio

The use of turbo codes enhances the data transmission efficiency and optimizes the performance of a communication system over wireless fading channels. In this paper, we present a brief overview of the various components of the turbo coding scheme, analyze the complexities of the most popular turbo decoding algorithms, and discuss the various implementation methods of the maximum a posteriori (MAP) algorithm. The paper considers the well-known log-MAP decoding algorithm by a linear approximation of the correction function used by the max* operator. We propose a generalized decoding scheme that optimizes the existing MAP algorithm for faster convergence and better throughput on the basis of varying channel conditions. The proposed scheme of decoding reduces complexity and enhances the throughput with only a negligible loss in BER performance.

A Fast Frequency-Domain Algorithm for Equalizing Acoustic Impulse Responses

The multiple-input/output inverse theorem (MINT) algorithm for multichannel equalization is computationally demanding. Although adaptive MINT reduces the computational complexity, it suffers from slow convergence. In this letter, we propose a low-complexity fast-converging adaptive algorithm for multichannel equalization. The novelty of the approach lies in the adaptive equalization for each frequency bin and its ability to achieve fast convergence in a single step. The proposed algorithm can achieve better equalization of high-order acoustic impulse responses with significant reduction in complexity.

Global optimization by small-world optimization algorithm based on social relationship network

A fast global convergence algorithm, small-world optimization (SWO), was designed to solve the global optimization problems, which was inspired from small-world theory and six degrees of separation principle in sociology. Firstly, the solution space was organized into a small-world network model based on social relationship network. Secondly, a simple search strategy was adopted to navigate into this network in order to realize the optimization. In SWO, the two operators for searching the short-range contacts and long-range contacts in small-world network were corresponding to the exploitation and exploration, which have been revealed as the common features in many intelligent algorithms. The proposed algorithm was validated via popular benchmark functions and engineering problems. And also the impacts of parameters were studied. The simulation results indicate that because of the small-world theory, it is suitable for heuristic methods to search targets efficiently in this constructed small-world network model. It is not easy for each test mail to fall into a local trap by shifting into two mapping spaces in order to accelerate the convergence speed. Compared with some classical algorithms, SWO is inherited with optimal features and outstanding in convergence speed. Thus, the algorithm can be considered as a good alternative to solve global optimization problems.

Design of a fast convergent backpropagation algorithm based on optimal control theory

The main contribution of this paper is using optimal control theory for improving the convergence rate of backpropagation algorithm. In the proposed approach, the learning algorithm of backpropagation is modeled as a minimum time control problem in which the step-size of its learning factor is considered as the input of this model. In contrast to the traditional backpropagation, learning algorithms which select the step-size by trial and error, it is selected adaptively based on optimal control criterion. The effectiveness of the proposed algorithm is evaluated in two simulations: XOR and 3-bit parity. In both simulation examples, the proposed algorithm outperforms well in speed and the ability to escape from local minima.

Extrapolation of Bandlimited Signals in Linear Canonical Transform Domain

The linear canonical transform (LCT) has been shown to be a powerful analyzing tool in signal processing. Many results of this transform are already known, including bandlimited extrapolation. The existing algorithm for solving the problem of LCT bandlimited extrapolation is based on signal expansion into a series of generalized prolate spheroidal wave functions (GPSWFs). However, the requirement to compute and store the GPSWFs and the errors due to the series truncation render this algorithm ill-suitable for a practical implementation. In this correspondence, we first propose a new formulation of the Gerchberg-Papoulis (GP) algorithm for LCT bandlimited extrapolation. Then, we present a fast convergence algorithm for the new formulation. The classical GP algorithm related to Fourier bandlimited signals is noted as a special case. Moreover, the comparison between the proposed algorithm and the one based on GPSWFs expansion is provided, and the validity of the theoretical derivations is demonstrated via simulations. Several potential applications of the achieved theory are also presented.

An Improved Gauss-Newtons Method based Back-propagation Algorithm for Fast Convergence

The present work deals with an improved back-propagation algorithm based on Gauss-Newton numerical optimization method for fast convergence. The steepest descent method is used for the back-propagation. The algorithm is tested using various datasets and compared with the steepest descent back-propagation algorithm. In the system, optimization is carried out using multilayer neural network. The efficacy of the proposed method is observed during the training period as it converges quickly for the dataset used in test. The requirement of memory for computing the steps of algorithm is also analyzed.

Fast Convergence Algorithm for Earthquake Prediction Using SLF/ELF Horizontal Electric Dipole during Day and Night and Schumann Resonance

Electromagnetic wave radiation from a SLF/ELF horizontal electric dipole (HED) related to seismic activity is discussed. In order to estimate the effects on the electromagnetic waves associated with the seismic activity, SLF/ELF waves on the ground radiated from a possible seismic current source modeled as a electric dipole, are precisely computed by using a speeding numerical convergence algorithm. A theoretical calculation of the VLF/SLF electric wave propagating among the Earth-ionosphere cavity generally utilizes the full wave method to solve the model equation. The field in the cavity is comprehended as the sum of each wave mode. However, this method is very complex, and unsuitable to the ELF frequency band. In 1999, Barrick proposed an algorithm, which was only suitable to solve the electromagnetic problems under the ideal electric conductor condition. To solve the problems under the non-ideal electric conductor condition, we have further developed Barrick's method and proposed a speeding numerical convergence algorithm. The spherical harmonic series expressions of electromagnetic fields excited by SLF/ELF HED in non-ideal Earth-ionosphere cavity are derived. The speed of this algorithm is faster thirty times than it of calculating directly the sum of the series. If it calculates directly the sum of the series, it needs 1,000 series items, while it needs only 200 series items by this algorithm. Our algorithm is compared with the second order spherical surface approximate algorithm, and two algorithms agree with each other very well. Therefore, our algorithm is correct. Schumann resonance is also verified.

The performance of monotonic and new non-monotonic gradient ascent reconstruction algorithms for high-resolution neuroreceptor PET imaging

Iterative expectation maximization (EM) techniques have been extensively used to solve maximum likelihood (ML) problems in positron emission tomography (PET) image reconstruction. Although EM methods offer a robust approach to solving ML problems, they usually suffer from slow convergence rates. The ordered subsets EM (OSEM) algorithm provides significant improvements in the convergence rate, but it can cycle between estimates converging towards the ML solution of each subset. In contrast, gradient-based methods, such as the recently proposed non-monotonic maximum likelihood (NMML) and the more established preconditioned conjugate gradient (PCG), offer a globally convergent, yet equally fast, alternative to OSEM. Reported results showed that NMML provides faster convergence compared to OSEM; however, it has never been compared to other fast gradient-based methods, like PCG. Therefore, in this work we evaluate the performance of two gradient-based methods (NMML and PCG) and investigate their potential as an alternative to the fast and widely used OSEM. All algorithms were evaluated using 2D simulations, as well as a single [11C]DASB clinical brain dataset. Results on simulated 2D data show that both PCG and NMML achieve orders of magnitude faster convergence to the ML solution compared to MLEM and exhibit comparable performance to OSEM. Equally fast performance is observed between OSEM and PCG for clinical 3D data, but NMML seems to perform poorly. However, with the addition of a preconditioner term to the gradient direction, the convergence behaviour of NMML can be substantially improved. Although PCG is a fast convergent algorithm, the use of a (bent) line search increases the complexity of the implementation, as well as the computational time involved per iteration. Contrary to previous reports, NMML offers no clear advantage over OSEM or PCG, for noisy PET data. Therefore, we conclude that there is little evidence to replace OSEM as the algorithm of choice for many applications, especially given that in practice convergence is often not desired for algorithms seeking ML estimates.

The Lee–Seo model with regularization term for bimodal image segmentation

Abstract In this paper, we improve Lee–Seo’ bimodal image segmentation model using a regularization term. This regularization term will maintain the smoothness of the level set function and decrease the level set function’ oscillations around the desired steady state when the noise level is lager. Furthermore, we also provide a rigorous study of the modified model. Based on techniques in calculus of variations, the existence of solutions of the modified model in BV space is established. Based on the theory we present (see Lemmas 2 and 3), we constructed a fast convergent algorithm to process images. It turns out our method is twice fast in processing an image than Lee–Seo’s algorithm with the same constant value initial level set function.

Modified Fourier-transform method for phase-shift calibration

Abstract A fast convergent algorithm to accurately measure phase-shift for phase shifter is proposed. In the presence of a spatial carrier frequency, an error function in the Fourier-domain is introduced and then the unknown phase-shift between two phase-shifted fringe patterns is determined by minimization of the error function with the help of a golden section search approach. Compared with the conventional Fourier-transform method, the modified method proposed in the paper is more immune to random noise and hence can give a more accurate result. The translation curve of a liquid-crystal spatial light modulator is also experimentally calibrated.

Generalized GIPSCAL re-revisited: a fast convergent algorithm with acceleration by the minimal polynomial extrapolation

Generalized GIPSCAL, like DEDICOM, is a model for the analysis of square asymmetric tables. It is a special case of DEDICOM, but unlike DEDICOM, it ensures the nonnegative definiteness (nnd) of the model matrix, thereby allowing a spatial representation of the asymmetric relationships among "objects". A fast convergent algorithm was developed for GIPSCAL with acceleration by the minimal polynomial extrapolation. The proposed algorithm was compared with Trendafilov's algorithm in computational speed. The basic algorithm has been adapted to various extensions of GIPSCAL, including off-diagonal DEDICOM/GIPSCAL, and three-way GIPSCAL.

Sonar target detection by modified adaptive noise cancellation using correlating filter

In this article, a new approach for active sonar-target detection based on a modified version of the method of adaptive noise cancellation (ANC) originally proposed by Widrow et al. (1975) is presented. Widrow's method requires a good estimate of reference noise in order to effectively ‘cancel’ the noise. A novel method of securing the required reference noise estimates is proposed in this article. The estimated reference noise should be highly correlated with the actual noise for effective cancellation of noise. In order to render this property, a correlating filter, designed according to a suitable optimality criterion, is made use of at the input of adaptive noise canceller. A new fast convergent adaptive algorithm is derived for updating the weights of the adaptive filter in the adaptive noise canceller. Target detection is effected by comparing the magnitude of the output of the adaptive noise canceller with a decision threshold. Computer simulations are carried out to demonstrate the performance of the proposed modified ANC-based method for target detection. The simulation results demonstrate that the proposed scheme is capable of performing target detection and signal-arrival time estimation with a very high degree of reliability at SNRs as low as −70 dB.