Traffic state prediction method based on improved SVM algorithm

Abstract

The invention provides a traffic state prediction method based on an improved SVM algorithm. The method comprises the specific steps that a historical traffic data sample set is preprocessed, sample data obtained after preprocessed is partitioned into k disjoint subsamples, one subsample is selected from the k disjoint subsamples to serve as a test dataset, and the remaining (k-1) subsamples serveas a training dataset; an improved SVM model is used to perform model training on the training dataset, and the improved SVM model is continuously optimized to generate an optimal improved SVM prediction model; and the test dataset is input into the optimal improved SVM prediction model, so that a prediction result is obtained. According to the method, the influence of an abnormal value in the data can be effectively relieved by use of the improved SVM algorithm, the robustness of the model is enhanced, the generalization capability of the model is improved, and prediction precision is improved.

Description

technical field [0001] The invention relates to the field of intelligent traffic system—city traffic state prediction, in particular to a traffic state prediction method based on an improved SVM algorithm. Background technique [0002] With the rapid development of society, the contradiction between supply and demand between limited road resources and the rapidly increasing number of motor vehicles has become more and more acute, resulting in a difficult balance between traffic supply and demand, which directly leads to increasingly severe traffic congestion in cities . Road traffic control and guidance system is an important method to deal with urban traffic problems, and short-term traffic state prediction is one of the key issues to realize the above methods. Therefore, in recent decades, researchers in this field have proposed various types of short-term traffic flow forecasting methods, including: short-term traffic flow combined forecasting method based on wavelet pac...

AI Technical Summary

Problems solved by technology

The traffic data collected from the scene, due to the influence of the measurement tools and the measurement environment, the collected traffic data inevitably contains abnormal values, and sometimes even contains serious errors

At present, the collected traffic data is only processed as follows: data cleaning (completion of missing data, identification of wrong data, reduction of redundant data), data standardization (summation standardization, standard deviation standardization, minimum and maximum value standardization, and extreme difference standardization) ), etc., it is prone to errors of discarding the true or taking the false, which directly affects the generated prediction model

Images