The Application Research Of Books Intelligent Distributing Based On BP Neural Network

The utilization of the books is the key to decide the quality of library. Recently, how to improve the utilization of the books has caused more and more attention. In order to solve the problem, valuable suggestions from many literatures have been proposed, such as the manage system, the reader's demand, picking out the old books and books distribution and so on. Based on the books distribution theory, the dissertation unified the BP neural network technology to propose design scheme of the intellectualized books distribution system, with the purpose to realize the books intelligent distribution, achieve the most superior result of the structure of the books distribution, and improve the utilization of the books.The application of the books intelligent distribution by is researched in this dissertation using BP neural network. The main contents conclude the process of features quantification and the algorithm based on the classification-learning of BP neural networks. Firstly, the data of books were chosen according to requirement of the question, and the features extracted with the principle of feature extraction, quantification processing, normalization processing and the data preprocessing were carried out, and the quantification features were transformed to the input signal for the neural network. Secondly, a 3-layer BP neural network with single hidden layer was designed, and a strategy of net-reconstructing and relearning was proposed. The adoptions of the strategy determine not only the nodes number of the best hidden units, but also the best neural network. Finally, combining BP neural network model with content of the dissertation, the way of classification learning to the data to create neural network and the classified learning algorithm based on BP neural network were proposed. Also the method was proved rationality, reliability and effectiveness with the k-fold cross-validation algorithm.The experimental results show that the system has solved the problem of books intelligent distribution, and further verified from the point of view of application the superiority of the neural network in solving problems of pattern recognition. Furthermore, it developed a helpful and technical exploration for intelligent management of library.