Advanced Materials Research Vols. 998-999

Abstract: The pattern recognition process control diagram, this paper puts forward a new method of training neural network. It only needs a small training data set can complete this work. This method is also compatible with the training algorithm, and get a better network performance. Pattern recognition success rate is very high in the larger parameter range, but also has some comparability.

Abstract: On the basis of analysis in Zhejiang, this paper uses the trend fitting method, the quadratic exponential smoothing model and multiple linear regression and grey GM (1, 1) portfolio model to forecast electricity consumption in 2012-2020 in Zhejiang, and compare the various methods of prediction accuracy.

Abstract: It combines the own characteristics of vehicle scheduling, taking into account the both interests of the bus company and passengers, and establishes the vehicle scheduling plan model, taken the departure time (real value) as the gene variables to encode, and taken the difference between the adjacent grid spacing, the departure time interval of maximum and minimum, passenger load factor etc. to limit. With the example, the model is improved to optimize the solution. Finally, it is possible to reliably find an optimal or near-optimal solution program on vehicle scheduling optimization problem in transit huge search space. Through comparative analysis of the experimental results, it can verify the effectiveness of the improved algorithm and superiority.

Abstract: In order to improve the efficiency of intelligence algorithm in integration planning and to avoid “premature” phenomenon, an improved genetic algorithm strategy-a method based on fitness density and scaling is proposed in the paper. It applies a new encoding way of integration planning. The two examples of simulation experiments show that the genetic improvement strategy could significantly reduce solving algebra and prevent “premature” phenomenon.

Abstract: With the fast development of vision-based hand gesture recognition, it is possible to apply the technology to astronaut virtual training. In order to solve problems of hand gesture recognition in future virtual training and to provide an unrestricted natural training for astronauts, this paper proposed a vision-based hand gesture recognition method, and implemented a hierarchical gesture recognition system to provide a gesture-driven interactive interface for astronaut virtual training system. The experiment results showed that this recognition system can be used to help astronaut training.

Abstract: Axiomatic Design (AD) provides designers with two axioms as criteria for design decision-making. The integration of AD with Theory of Inventive Problem Solving (TRIZ) is expected to effectively improve design activities. A computer-supported axiomatic design system integrating TRIZ methods was proposed to help designers obtain innovative solutions and make right decisions. Within the AD framework, TRIZ not only can be applied to adjust design parameters when the design is coupled, but also be used to describe functional requirements and directly obtain proper solutions. A fuzzy design scheme evaluation method was presented to improve the implementation of the information axiom. Finally, a computer aided design prototype system based on the proposed model was developed. And a case study of the structure improvement of a faucet demonstrated the validity and practicability of the system.

Abstract: Aiming at solving the problem of forecasting the dynamics of express delivering quality elements classification, the paper proposed a novel method integrating Kano model with GM(1,1) model which is effective to perform fitting prediction. An approach of fuzzy questionnaires to modify Kano questionnaires which considered as subjective is suggested, and the GM (1,1) model is used to estimate and predict data according to the quality classification of fuzzy Kano model. An empirical study in express industry is provided to testify its feasibility and validity.

Abstract: Through nation-wide and various provinces and regions of the country's gross national product and software industry data Investigation. Combined with industrial concentration index method and location entropy method on the cluster development of software industry and software industry in Jilin province has carried on the empirical research, analysis the software industry in Jilin province while keeping the sustainable development. But the agglomeration development is weak. The software business income, agglomeration degree and the speed of converging in the software industry are below the national average.

Abstract: Based on the modified grey prediction model, the outputs of software industry in Jilin Province were predicted. First the historical data and updated the data were pre-treated by iteration. Then it was found that the results from the modified grey prediction model were better than that from traditional grey prediction model by residual analysis. Finally, the prediction about the outputs of software industry in Jilin Province was given for the next five years. According to the experimental results, our proposed new method obviously can improve the prediction accuracy of the original grey model.

Abstract: Feature modeling is a main stream technology in domain requirement analysis of software product line engineering. Establishing the traceability between feature model and software architecture plays the essential role in improving software quality. Based on Formal Concept Analysis technology, we proposed a method to verify traceability between feature model and software architecture. The method analyzed the constitution of the feature, defined feature model and software architecture function expression, constructed a concept lattice and presented several mapping criteria to analysis it. It is more applicable for higher complexity model, comparing to the other model traceability method. A case study is used to demonstrate the feasibility of the method.