Papers by Keyword: Conjugate Gradient Method

Abstract: We developed a parallelized version of a numerical simulator for flow in shale gas reservoirs. For this purpose, the OpenACC Application Programming Interface (API) was used to parallelize specific parts of the original serial code, enabling its execution in multiple cores. The study addressed the simulation of production considering the use of a vertical well and the effects of adsorption and slippage in a two-dimensional domain in Cartesian geometry. For the discretization of the equations that govern the flow, we used the Control Volume-Finite Difference method and the Conjugated Gradient method to obtain the solution for the pressure field. In addition, we carried out a sensibility analysis varying the computational mesh and the number of threads. The results obtained showed that there were significant gains in terms of computational performance.

Abstract: In this paper, we introduce an algorithm for solving large-scale box-constrained optimization problems. At each iteration of the proposed algorithm, we first estimate the active set by means of an active set identification technique. The components of the search direction corresponding to the active set are simply defined; the other components are determined by nonlinear conjugate gradient method. Under some additional conditions, we show that the algorithm converges globally. We also report some preliminary numerical experiments to show that the proposed algorithm is practicable and effective for the test problems.

Abstract: On account of low convergence of the traditional genetic algorithm in the late,a hybrid genetic algorithm based on conjugate gradient method and genetic algorithm is proposed.This hybrid algorithm takes advantage of Conjugate Gradient’s certainty, but also the use of genetic algorithms in order to avoid falling into local optimum, so it can quickly converge to the exact global optimal solution. Using Two test functions for testing, shows that performance of this hybrid genetic algorithm is better than single conjugate gradient method and genetic algorithm and have achieved good results.

Abstract: As a typical reciprocating engine power machinery, complex structure determines its failure brings about the complexity and diversity, it shows the uncertainties of operating environment, system noise and sensor accuracy, and engine fault diagnosis accuracy rate is reduced, taking into account the limitations of traditional BP neural networks, improved BP algorithms include statistical algorithms, additional momentum method, variable learning rate method and conjugate gradient method are studied. Finally, the engine is as an example, engine fault diagnosis experimental system is set, the vibration signals are measured in the normal state, left one and right six cylinders off the oil and air filter blockage in the load of 2565Nm, and the speed of 1500r/min, 1800r/min and 2200r/min. The test and analysis by comparing above mentioned methods indicate it is verified the superiority improved BP neural network with the conjugate gradient method.

Abstract: The steel tubular arch is hoisted segment by segment through cable crane and the stayed cables are used to maintain stability and balance. The determination of the stayed-cable forces and construction camber value of the erected rib segments becomes the key issue to ensure construction quality and safety. The forward iteration analysis method, which combines finite element method with optimization method as provided in this paper, can easily and effectively determine the stayed-cable forces and construction camber value in the erection of the rib segments, and the stay cables can be tensioned to their target force values only at one time. Finally, the example is demonstrated to prove the correctness and affectivity of the present method. Numerical example indicates that the results based on the method may be used to the backward analysis of the initial state, and that the proposed CM has excellent features of quick convergence rate and best global performance.

Abstract: This paper proposes the combined direction stochastic approximation method for solving simulation-based optimization problems. The new algorithm is a stochastic analogy of conjugate gradient method, which employs a weighted combination of the current noisy negative gradient and some former noisy negative gradient as iterative direction. Our numerical experiments show that the new algorithm outperforms the classical RM algorithm for two typical simulation-based optimization problems, a.e., M/M/1 queuing problem and inventory problem.

Abstract: This paper mainly researches the inverse problems of the electrical exploration results. Through the inverse calculation using genetic algorithm and conjugate gradient method, the model parameters and inverse curves can be obtained. In the paper, the exploration abnormal results of apparent resistivity are inverse analyzed under the anomaly condition of double-pole device of electrical exploration. The inverse results of the model and theoretical analysis are compared. The result indicates that genetic algorithm and conjugate gradient method can be used to obtain the inverse parameters of the exploration results, but the precision of the parameters is different in different geological conditions and methods.

Abstract: A modifying Sub-threshold Weak Conjugate Gradient Pursuit Method is proposed to improve recovery accuracy of Compressed Sensing. This method adds a direction in the Directional Pursuit Method and uses a stopping criterion for the indices of elements, then the evaluation of sparse signal is obtained by using Least Square Method. The simulation shows that for the same sparsity level, the number of measurements needed by the method is less than that needed by MP or StOMP-FDR to exactly recover, and the recovery accuracy is higher.

Abstract: The compressive stress distribution below the specimen surface of a nanocrystalline medium carbon steel was investigated nondestructively by using high-energy X-rays from a synchrotron radiation source, SPring-8 (Super Photon ring-8 GeV) in the Japan Synchrotron Radiation Research Institute. A medium carbon steel plate was shot-peened with fine cast iron particles of the size of 50 μm. By using the monochromatic X-ray beam with three energy levels of 10, 30 and 72 keV, the stress values at the arbitrary depth were measured by the constant penetration depth method. The stress was calculated from the slope of the sin2ψ diagram. Measured stress corresponds to the weighted average associated with the attenuation of the X-rays in the material. The real stress distribution was estimated by using the optimization technique. The stress distribution was assumed by the third order polynomial in the near surface layer and the second order polynomial. The coefficients of the polynomials were determined by the conjugate gradient iteration. The predicted stress distribution agreed well with that measured by the conventional surface removal method.