Papers by Keyword: Evolutionary Neural Network

Abstract: It is an important work for modern libraries to predict reader flow. With the help of reader flow, library staff can grasp the change regulation of readers, allocate tasks rationally and take steps ahead of time in high-risk period. Because of reader flows typical non-linear characteristics, evolutionary neural network technology is introduced in this research so as to improve the accuracy of reader flow prediction. A prediction method for library reader flow based on evolutionary neural network is proposed. Genetic algorithm is used to optimize and design BP neural network firstly, then evolutionary neural network is used to predict reader flow. The experimental results show that evolutionary neural network is an effective tool for us to predict library reader flow. We can realize an accurate prediction for library reader flow by this method.

Abstract: Determination of sliding body saturated permeability coefficient has an important significance for groundwater dynamic field, deformation and stability of reservoir landslide. Based on in-depth analysis of landslide geological characteristics, groundwater monitoring data, rainfall and reservoir water lever in Three Gorges Reservoir, it is found that reservoir water level is the main factor which affecting the groundwater dynamic field. According to two layers of sliding body, the saturated permeability coefficient test plan of two factors and three levels has been designed. Based on saturated-unsaturated seepage finite element and the reservoir water level fluctuation curve, groundwater saturation line has obtained under different combinations of parameters. Combined with genetic algorithm, an optimal neural network model was established to describe the nonlinear relationship between the sliding body saturated permeability coefficient and borehole water level. Then based on the monitoring data of borehole water level, the saturated permeability coefficient of two layers sliding body can be gotten. Finally, with the inversion parameters, it is found that the calculating water level results and the actual level of underground water are in good agreement. Results show that the inverse thought and method is feasible and effective, which can provide reference for the similar engineering.

Abstract: Data classification is the foundation for the intelligent identification and management of massive information in the internet of things. To classify the massive data accurately, an evolutionary neural network is presented. The input features and the structure of neural network are evolved simultaneously to consider their joint contribution to the performance of neural network. The sensitivity analysis is performed to guide the evolutionary algorithm to search the optimum solution. It can be seen from the experimental results that the proposed evolutionary algorithm optimized the structure of neural network and eliminate the tedious input features at the same time. The excellent classification accuracy is achieved finally.

Abstract: In this study, a genetic algorithm simulating human reproduction mode (HRGA) is proposed. The genetic operators of HRGA include selection operator, help operator, crossover operator and mutation operator. The sex feature, age feature and consanguinity feature of genetic individuals are considered. Two individuals with opposite sex can reproduce the next generation if they are distant consanguinity individuals and their age is allowable. Based on this genetic algorithm, an improved evolutionary neural network algorithm named HRGA-BP algorithm is formed. In HRGA-BP algorithm, HRGA is used firstly to evolve and design the structure, the initial weights and thresholds, the training ratio and momentum factor of neural network roundly. Then, training samples are used to search for the optimal solution by the evolutionary neural network. HRGA-BP algorithm is used in motor fault diagnosis. The illustrational results show that HRGA-BP algorithm is better than traditional neural network algorithms in both speed and precision of convergence, and its validity in fault diagnosis is proved.

Abstract: The evolutionary neural network can be generated combining the evolutionary optimization algorithm and neural network. Based on analysis of shortcomings of previously proposed evolutionary neural networks, combining the continuous ant colony optimization proposed by author and BP neural network, a new evolutionary neural network whose architecture and connection weights evolve simultaneously is proposed. At last, through the typical XOR problem, the new evolutionary neural network is compared and analyzed with BP neural network and traditional evolutionary neural networks based on genetic algorithm and evolutionary programming. The computing results show that the precision and efficiency of the new neural network are all better.

Abstract: Assembler Encoding is Artificial Neural Network encoding method. To date Assembler Encoding has been tested in the optimization problem and in the so-called predator-prey problem. The paper reports experiments in a next test problem, i.e. in the inverted pendulum problem. During the experiments two direct encodings were also tested in order to compare Assembler Encoding with other Artificial Neural Network encoding methods.

Abstract: This paper reports a systematic investigation of high speed grinding of hard-to-machining of titanium alloys. The ground surfaces were characterized using scanning electron microscopy, and the effects of different grinding parameters on roughness were discussed. A numerical model was established to predict surface roughness based on the evolutionary neural network optimized by Genetic Algorithm (GA). The modeled results were in good agreement with the experimental results.