Journal of Computational Science & Engineering

 

 

 

 

 

ISSN 1710-4068                ACSS home     Journal's home     

    J. Comput. Sci. Eng.  Vol. 11 (2014) 413-437
 

Dynamic System Method for Solving Inverse Problems in Heat Conduction Equations

 
 
Junli Zhang and Tingting Sheng

   
J. Comput. Sci. Eng. 11 (2014) 413-417 !Published March 21, 2014    
 

Abstract:Dynamic system method is an effective method for solving algebraic system. It is not only be applied to well-posed operator equation but also is suitable for nonlinear and ill-posed operator equations. In this paper, dynamic system method is used to solve the inverse problem of heat conduction equations, and competes with classic Tikhonov regularization algorithms. The numerical tests demonstrate dynamic system method is a fast and simply algorithms.

   
Keywords:Dynamical Systems Methods; Inverse problems; Heat Conduction Equations; Stability.

: Download (free)  CSE-PDF

 

Quantitative structure-activity relationship study on toxicity to tetrahymena pyriformis of aniline derivatives

 
 
Tianyang Gao    

J. Comput. Sci. Eng. 11 (2014) 3418-422  ! Published April 20, 2014

   
 
Abstract: A quantitative structure-activity relationship was developed by heuristic method (HM) and radial basis function neural network (RBFNN) to study the tetrahymena pyriformis toxicity (�ClogIGC50) of 48 aniline compounds. The statistical parameters provided by the HM model were R2 = 0.934; F = 159.96; RMS = 0.1761 for the training set and R2 = 0.894; F = 67.309; RMS = 0.2613 for the test set. The non-linear RBFNN model gave a better result R2 = 0.968; F = 1087.692; RMS = 0.1159 for the training set and R2 = 0.910; F = 80.905; RMS = 0.2106 for the test set. Cross-validation was used to evaluate both models.
   
Keywords:Anilines; Quantitative structure-property relationship (QSPR); Heuristic method (HM); Radial basis function neural network (RBFNN).

: Download (free)  CSE-PDF

 

Load forecasting utilizing an improved Wavelet Neural Network (WNN) model

 
 

Qingxiang Ou

   

J. Comput. Sci. Eng. 11 (2014) 423-427  ! Published April 21, 2014

   
 
Abstract:Short-term load forecasting is for the power system and has great significance for the operation stable stir and development of the power system. Wavelet neural network model is an effective method to solve the problem of power system load forecasting. In this paper, increasing momentum term was employed to improve the traditional wavelet neural network model. Increasing momentum term could enhance learning efficiency of network and propose an improved wavelet neural network model which was applied to short-term load forecasting in a certain area of Beijing. The results show that the improved Wavelet neural network model has a better prediction performances and more suitable for short-term load forecasting, compared with the traditiona model. 
   
Keywords:Load forecasting; Wavelet neural network; Momentum term; Prediction performance.
 

: Download (free)  CSE-PDF

 

Page start 328 A Survey on Selfishness Incentive Mechanism in Opportunistic Networks

 
 

Li Liu

   

J. Comput. Sci. Eng. 11 (2014) 428-431  ! Published April 21, 2014

   
 
Abstract:Opportunistic Networks enable mobile devices to communicate in an environment where contemporaneous end-to-end paths are unavailable or unstable. Message delivery in Opportunistic Networks is implemented through store carry and forward fashion by use of short message transmission technologies such as Bluetooth or WiFi. To stimulate cooperation and prevent selfish behavior is an important challenge in Opportunistic Networks. This paper surveys the recent incentive mechanisms in Opportunistic Networks.
   
Keywords: DTNs; Incentive Mechanism; Selfishness.
 

: Download (free)  CSE-PDF

 

Fuzzy GM (1, 1) Model and Application in Building Settlement Analysis

 

Huizhou Li

 

J. Comput. Sci. Eng. 11 (2014) 432-437 ! Published April 21, 2014

  Abstract:In order to improve the prediction accuracy, GM (1, 1) model is improved, and a fuzzy GM (1, 1) prediction model is presented in this paper, named as FGM (1, 1). In this method, a fuzzy function is introduced into GM (1, 1) mode, by which the data of time series is fuzzed to optimize the selection of data, and the forecasting of history data >fart her is more weight >can be achieved. Building settlement analysis results with the application of FGM (full name) (1, 1) show that this forecast method is effective and reliable, which offers new approach to improving forecast accuracy.
  Keywords:Fuzzy GM (1, 1); Fuzzy function; Settlement analysis.
 

: Download (free)  CSE-PDF

!! from ACSS

 

About Us | Site Map | Privacy Policy | Contact Us |    Copyright © 2004- 2013 The American Computational Science Society