Applied Mechanics and Materials Vols. 380-384

Abstract: As a result of poor local research performance and short of the population diversity in niche genetic algorithm (NGA), an improved niche genetic algorithm (I-NGA) is presented in this paper. The algorithm adopts migration strategy, grading gradient method and gene equilibrium strategy, and it has better performance on accelerating the convergence rate and increasing the population diversity. This approach is used in Camel function for confirming. Through comparisons to NGA and I-NGA, the improved algorithm shows its better optimization.

Abstract: We present a new least square method called Symmetrical Least Square Method (SLSM) for linear fitting. It is symmetrical to two groups of fitted trial data. With SLSM, the slope and its uncertainty of a fitting line are given on zero-crossing linear fitting. The preliminary application of the SLSM supports that it has less uncertainty than routine least square method.

Abstract: The Traveling Salesman Problem is one of the most intensively studied problems in computational mathematics. Due to the basic genetic algorithm convergence speed is slow, easy to stagnation. We present in this paper a new improved mutation strategy to solve this problem. Roulette wheel selection strategy is used to avoid running into trap of the part best value. And simulated niche method is introduced to accelerate the search process effetively. This algorithm has been checked on a set of 144 cities in China and it outperforms the results obtained with other TSP heuristic methods.

Abstract: The readers satisfaction rate is an important index of library service quality. Evaluation of library readers satisfaction rate depends on knowledge rules to a large extent. In this study, the synthesized evaluation index system of readers satisfaction rate was established firstly. Then, a method for mining the evaluation knowledge rules of readers satisfaction rate based on an improved genetic algorithm was proposed. In the algorithm, new knowledge rules were generated by selection operator, dual crossover operator and dual mutation operator. Knowledge rules were evaluated by their accuracy, coverage and reliability, and they were evaluated by linear combination method. Experimental results show that this method for mining knowledge rules is valid. It is helpful for us to evaluate library readers satisfaction rate fairly and objectively.

Abstract: In order to realize the adaptive Genetic Algorithms to balance the contradiction between algorithm convergence rate and algorithm accuracy for automatic generation of software testing cases, improved Genetic Algorithms is proposed for different aspects. Orthogonal method and Equivalence partitioning are employed together to make the initial testing population more effective with more reasonable coverage; Genetic operators of Crossover and Mutation is defined adaptively by the dynamic adjustment according to multi-objective Fitness function, which can guide the testing process more properly and realize the biggest testing coverage to find more defects as far as possible. Finally, the improved Genetic Algorithm are compared and analyzed by testing one benchmark program to verify its feasibility and effectiveness.

Abstract: Partition methods for real data play an extremely important role in decision tree algorithms in data mining and machine learning because the decision tree algorithms require that the values of attributes are discrete. In this paper, we propose a novel partition method for real data in decision tree using statistical criterion. This method constructs a statistical criterion to find accurate merging intervals. In addition, we present a heuristic partition algorithm to achieve a desired partition result with the aim to improve the performance of decision tree algorithms. Empirical experiments on UCI real data show that the new algorithm generates a better partition scheme that improves the classification accuracy of C4.5 decision tree than existing algorithms.

Abstract: For a long time, the mathematical model is used to describe the systems characteristics and acquire the solutions, so it gradually develops into a modern computer simulation technology, and it can be used to solve many problems which are complex and unable resolve use mathematical methods. Witness modeling simulation software is applied in this article, and examples analysis is put forward in view of the logistics system of inventory control, finally making the models on the part of the module, whats more, it analysis the data and makes the optimal strategy.

Abstract: in order to improve the efficiency of ray-space data compression, according to the texture characteristics of ray space data, introducing a simple effective method of texture classification, a new fast block matching algorithm based on adaptive template selection is proposed in this paper, for prediction coding of ray space slice sequence. Experimental results show that the proposed algorithm has low-complexity and high-performance characteristics, for different types of ray space slice sequences with strong adaptability.

Abstract: Multi-robot path planning using shared resources, easily conflict, prioritisation is the shared resource conflicts to resolve an important technology. This paper presents a learning classifier based on dynamic allocation of priority methods to improve the performance of the robot team. Individual robots learn to optimize their behaviors first, and then a high-level planner robot is introduced and trained to resolve conflicts by assigning priority. The novel approach is designed for Partially Observable Markov Decision Process environments. Simulation results show that the method used to solve the conflict in multi-robot path planning is effective and improve the capacity of multi-robot path planning.

Abstract: In the research and development of intelligence system, clustering analysis is a very important problem. According to the new direct clustering algorithm using similarity measure of Vague sets as evaluation criteria presented by paper, the Vague direct clustering method are used to analysis using different similarity measure of Vague sets. The experimental result shows that the direct clustering method based on the similarity of Vague sets is effective, and the direct clustering method based on different similarity measure of Vague sets is the same basically, but difference on the steps of clustering. To select different algorithms according different conditions in the work of the actual applications.