Research On Railway Freight Demand Forecasting Technology Under Uncertain Environment

Railway freight demand is an important part of the whole society's freight demand.Accurate prediction and grasp of freight demand and development trends is an important basis for railway companies to make development plans for railways.After decades of research by experts and scholars,a large number of excellent freight prediction algorithms have been produced.However,the generation and development of freight demand are closely related to many factors,and these factors are also constantly changing.In this uncertain environment,how to make more accurate forecasts of railway freight demand faces new problems and challenges.At present,the freight demand forecasting algorithm lacks effective means to study and analyze the group in all railway demand markets with limited modeling workload.Moreover,during the forecasting process,uncertain fluctuations such as random changes and sudden changes in freight demand due to environmental influences are ignored.In this paper,the research of freight demand forecasting technology under uncertain environment is carried out in response to the above problems.Two algorithms are proposed for freight demand forecasting.The specific research contents are as follows:(1)Faced with the problem of numerous diversified demand sides in the railway freight market,it is difficult to predict and analyze specifically,the clustering idea is innovatively introduced,and a freight demand clustering algorithm that meets the characteristics of freight transport is established.Affected by various factors of uncertain environment,the fluctuation of freight demand sequence has the characteristics of nonstationarity and nonlinearity,local analysis is carried out from the point of view of the time domain of the demand sequence.The complex and changeable freight demand sequence is decomposed into several components with different frequencies,and the component series are extracted into high frequency,low frequency and trend demand sequences by means of mean reconstruction to represent the local characteristics of the original freight demand.Combined with the k-means++ algorithm,the similarity distance is calculated for cluster analysis.In this paper,the validity of the algorithm is verified on the freight demand data set,and the characteristics of various demand changes are briefly analyzed.(2)In response to the uncertain fluctuations of the freight demand sequence,such as random changes and sudden changes.First of all,the qualitative and quantitative analysis of railway freight demand forecasting index is carried out.On the basis of the establishment of the index system,the freight demand forecasting model combining the nonlinear fitting ability of the neural network with the uncertainty theory is established.By performing uncertain processing on the output layer weights in the network,redefining the prediction objective function,and introducing uncertain amplitude and comprehensive uncertain amplitude definitions,a linear equation group for solving the prediction model is constructed and solved.Finally,a freight demand forecast model based on uncertain theory is established to solve the possible range of initial freight demand.Through the example verification of freight demand data,the results show that the model proposed in this paper is better than the existing models,which can better cope with the uncertainty changes of freight demand and improve the forecast accuracy.