Zhang Y.-X.,Beihua University |
Liu Y.,Jilin Railway Vocational and Technology College
Applied Mechanics and Materials | Year: 2014
Cloing and hypermutation of immune theory were used in optimization on particle swarm optimization (PSO), an immune particle swarm optimization (IPSO) algorithm was proposed, which overcome the problem of premature convergence on PSO. IPSO was used in BP Neural Network training to overcome slow convergence speed and easily getting into local dinky value of gradient descent algorithm. BP Neural Network trained by IPSO was used to fault diagnosis of power transformer, it has high accuracy after experimental verification and to meet the power transformer diagnosis engineering requirements. © (2014) Trans Tech Publications, Switzerland.
Zhang Y.,Beihua University |
Liu Y.,Jilin Railway Vocational and Technology College |
Duan H.,Beihua University
Advanced Materials Research | Year: 2013
The principle, ideas, procedures and optimization problems of the filtering algorithm were presented in the article. First, the filtering algorithm procedures of aerial image based on stochastic resonance theory was introduced, then the curve of noise strength and peak signal to noise ratio (PSNR) was analyzed, a golden-section fast search algorithm was proposed and discussed as well to find out the maximum point of PSNR in the process of filtering. Experimental results indicate that the filtering method in this article is superior to other methods on high noise strength aerial image, and the stability and robustness are better than other methods. The multiples of filtering image's PSNR is 1.2~1.4 times than other methods. © (2013) Trans Tech Publications, Switzerland.
Yanchun H.,Jilin Railway Vocational and Technology College
Sensors and Transducers | Year: 2013
Wireless sensor network localization involves the dataset of physical space and signal space, and its layout accord with the characteristic of manifold. This paper uses the own characteristic of topological structure and signal space,striving to search a right modeling approach to realize the mapping between these two spaces, thus the usage of locality preserving canonical correlation analysis (LPCCA)is put forward to realize the manifold structure and localization. To further improve the accuracy of localization algorithm, through combining specific positioning problem as regression rather than dimensionality reduction to adapt new application, developed an original wireless sensor network localization algorithm LE-LPCCA (location estimation-LPCCA) which can reflect network topological structure information. © 2013 IFSA.
Gao G.,Jilin Agricultural Science and Technology College |
Chen H.,Jilin Railway Vocational and Technology College
2010 International Conference on Computer, Mechatronics, Control and Electronic Engineering, CMCE 2010 | Year: 2010
ART-2 is a self-organized and unsupervised artificial neural network which can recognize complicated inputting patterns. This paper presents a new structure aiming to discard small amplitude information during classifying the data by Standard ART-2 network, especially the time series data. So, we proposed an improved structure based on standard ART-2. Finally, the problem of "discard small amplitude information" is successfully resolved by improved structure and a simulation is given to show the superiority. © 2010 IEEE.
Cui Y.-Y.,Jilin Railway Vocational and Technology College |
Zhang Z.,Shanghai University
Academic Journal of Second Military Medical University | Year: 2011
Objective To analyze the topological properties of genomic-wide genetic interaction network in yeast. Methods The topological properties of genomic-wide genetic interaction network in yeast were calculated by the graph theory. Results A power law fit the degree distribution well in the genomic-wide genetic interaction network in yeast, with the exponent approaching 3. TheNode's degree followed awídc-taíl distribution, with the average degree being 87. Two thirds of the genes had genetic interaction via only one gene, and about one third of the genes had genetic interaction via only 2 genes. The average clustering coefficient was 0. 047. Conclusion The gene function in yeast is usually multiple and the functional interaction between genes is highly condensed.