A Study Of Neural Networks In Forecasting Of Logistics System

Along with the enhancing of civil economy and the quickening of the international economic integration, the business between Cities, regions and countries expends so rapidly that technology of logistics forecasting is being paid more and more attention to. The commonly used quantitative forecasting method mainly includes the model of time series and cause-effect model, which need to set up corresponding mathematics model according to the historical materials and to makes prediction of the development trend of the logarithm row. Logistics system has sorts of complicated influent factors, and with the development of modern logistics, the data amount increase severely in which a lot of targets in fact are so complicated uncertainty to modeling or very difficult accurate modeling. This proposes the new direction of the research of the logistics forecasting technology.The Artificial Neural Network (abbreviated as ANN) rising in recent years has abilities such as showing random Nonlinear relations and studying etc., and offers the new thoughts and new methods to solve this kind of problem. The ANN research method has very strong robustness and fault-tolerance quality, and is good at associating, summarizing, analogy and reasoning, and have very strong since characteristics, learning ability, etc., the properties above all can give a help of analysis from a large amount of statistical data that draw from the macroscopical statistics materials. So, launching the forecasting study based on ANN has very good realistic meanings.This paper thoroughly and deeply predicts the theory and method of logistics forecasting based on the ANN researching. After synthetically introducing the development characteristics of modernlogistics and the current situation of the common-used forecasting theory applied in logistics system, the author points out the difficulty and insufficient may be met when the theory used in practical problem, and explains in detail that the research of the ANN forecasting theory and its applied situation. Then proceeding from request for the logistics projects, the paper proposes an improved 3-BP learning algorithm, and derives the basic principle of expounding this kind of improvement algorithm through mathematics. On this basis, the paper sets up tentatively an ANN forecasting model which combining many kinds of precision-evaluations and may accord with the modern logistics, and offers the detailed theory method appears and design-steps. After study of the combined forecasting methods based on the ANN theory, it is put foreword that Exponential-Smooth (ES) and ANN combine a new prediction algorithm. In order to observe the validity of method that the paper puts forward, the logistics forecasting system (LFS) is developed with the visual technology. The LFS system including many methods such as ES and Gray model, etc., can get the satisfied result while applied to the logistics project of the transportation.