Study Of Data Statistics Based On The Neural Network

In data age, it is a question to be faced to process large-scale data effectively. Data is a kind of source, and its value lies in the knowledge and information extracted from it and does not lie in itself. The data statistics was devoloped just to satisfy this requirement. Seen from the development history of data statistics, the early research mainly focuses on the traditional analysis method, and it is in this century ones find the data statistic can be done by intelligent methods. Furthermore, it has been verified by practice that to apply neural network to data statistics is an good method.Since neural network does not own good adaptive ability and learning ability, this thesis investigates the collection and analysis method of passenger flow volume based on the method of nueral network and the character of data collection by infrored equipment and designs a new people-counting system. According to the analysis and the natural relationn between input and ourput, an effectively neural network model is formulated, and expected output is obtained by using the effective input data. Applying the technology of neural network to the formulation of statistical model just utilizes its speciality in dealing with nonlinear problem.In the application of neural network classifier, this thesis solves the traffic statistics problem by using the improved BP neural network algorithm. Then, one optimizes the parameters in the model and computes the passenger flow volume by the effective connection of neural network and the counting method of passenger flow volume. The results of experiment show that the method based on neural network method not only can divide the data of passenger flow volume but also can identify the situation that many people go parallell, it has been illustrated that it is a kind of real-time method counting the passenger flow volume with high accuracy and efficiency.