Advanced Materials Research Vols. 655-657

Abstract: An improved genetic algorithm is applied to solve the problem of the robust optimization for structures under complicated loading. The objective function is constructed by signal-to-noise ratio of Taguchi target. Numerical examples demonstrated that the improved genetic algorithm combined with the robust design method can effectively solve the problem of the robust optimization for structures under complicated loading with uncertain parameters.

Abstract: The paper proposes a particle pair optimization (PPO) algorithm for numerical optimization. In the paper, a particle pair search(PPS) is designed. Based on PPS, the competitive roulette strategy is employed to the selection of particle pair from swarm, and a particle pair optimization algorithm is proposed. In the experiments, PPO is applied to solve the 12 benchmark problem, and compared with fast evolutionary programming (FEP). The results demonstrate that PPO algorithm can find optima or close-to-optimal solutions of the complex functions with high-dimension, and the search of PPO is stable and efficient.

Abstract: C4.5, Bayesian network and Sequential Minimal Optimization (SMO) are three typical classification algorithms in data mining. Using cross-validation method with 10 folds get analysis and calculation results of the experiments for the three classification algorithms in the same training set and test set. The main metrics include accuracy, precision, speed, robustness, scalability and comprehensibility, we use margin curve show these. It provides a theoretical and experimental basis for users to select a proper classification algorithm with different training sets in quality and amount.

Abstract: A novel image recognition method based on chaotic-particle swarm-optimization-neural network algorithm was presented. The chaotic mapping mechanism and particle swarm algorithm were used to optimize the weight and threshold of BP neural network which was applied to the recognition of image. The simulation results show this new method can overcome the problems that BP neural network is easy to fall into local optimum and sensitive to the initial value, and has better recognition rate and stronger robustness.

Abstract: In power quality monitoring system, there are unavoidably existing various kinds of noises in collected data，the presence of noise may result in increased false classification rate, denoising is an extremely important work for detection and classification of power quality disturbances. In order to improve the denoising result of power quality signal, an denoising method for power quality signal using Savitzky-Golay is proposed. Numerical results show that the proposed method can eliminate the influence of noise components and implement transient power quality disturbance detection and localization, thus providing good foundations for transient power quality disturbance monitoring under noise environment.

Abstract: The Time-Interleaved ADC（TIADC) is an effective method for implement ultra high-speed data acquisition. However, the errors of channel mismatch are seriously degrade the signal-to-noise ratio of the system, such as Time-skew error, Gain error and Offset error. This paper have done some researches and analysis, and given the modeling of the three channels mismatch. What's more, it also given a detailed analysis of error and the method of measure it, derived the formula of signal to noise and distortion ratio(SINAD) and spurious free dynamic range(SFDR). All of them provide a reference for the tolerance range of TIADC channel mismatch error. Meanwhile, the result of this paper has provided a theoretical basis for eliminating TIADC channel mismatch error.

Abstract: Focus on the problem of de-noising signals smoothness and similarity.The three signals were processed by four signal de-noising methods,which are inhibition detail coefficients,Fourier transform,global threshold and layered threshold method.And the energy ratio(PERF) and standard deviation(ERR) were obtained.Experiment results show that the global threshold de-noising method is the best for its high similarity;the layered threshold de-noising method is the best for its high smoothness.

Abstract: This paper proposed a novel method to segment the keyboard characters and symbols. An algorithm based on mathematical morphology has been applied to the keyboard inspection system. By using various morphology operations like opening and dilation, the algorithm overcomes the disadvantages of the traditional method such as time-consuming and unreliable. Therefore, our approach is capable of segmenting the characters and symbols accurately and we at last get the region of interest (ROI) to represent the keyboard characters and symbols.

Abstract: Visual inspection has been widely used in the field of automobile industry. On a car production line, we use an online monitoring system to measure the same feature region of a work piece. Traditionally one picture is chosen as the template and used for other target measurements. However, this method tends to causing measurement error. In this paper a new algorithm is proposed which can generate a better template by multiple pictures. The process is as following: we firstly match the feature area; secondly, the gray information of this area is segmented; then after optimizing a new template is finally formed. The experimental results show that the NMSD (Normalized Mean Square Difference) is 0.0047 when we use the old template to match 10 pictures, but with the new one, the NMSD is only 0.0005. These results show that the new template is more accurate and stable.

Abstract: SOM (Self-organizing Map) algorithm is a clustering method basing on non-supervision condition. The paper introduces an improved algorithm based on SOM neural network clustering. It proposes SOM’s basic theory on data clustering. For SOM’s practical problems in applications, the algorithm also improved the selection of initial weights and the scope of neighborhood parameters. Finally, the simulation results in Matlab prove that the improved clustering algorithm improve the correct rate and computational efficiency of data clustering and to make the convergence speed better.