Traffic flow predicating method based on sliding window average

Abstract

The invention relates to a traffic flow predicating method based on sliding window average. The traffic flow predicating method comprises the following steps that historical data and data on a predicating day are collected, a window value is set according to the data, then, model parameters are trained, and target data are worked out according to a traffic flow predicating model; next, the sliding is carried out for a certain interval, a predicating algorithm is invoked again for calculation until the predication of the data in all intervals is completed, and finally, the predicating average value of all the intervals is worked out. The traffic flow predicating method has the advantages that the irregular fluctuation of traffic flow data can be smoothened, the random fluctuation in dynamic data is eliminated, errors caused by data are reduced, and the like; the accuracy and the reliability of predicating results are improved.

Description

technical field

[0001] The invention relates to the field of intelligent traffic science, in particular to a traffic flow prediction method based on sliding window averaging. Background technique

[0002] With the progress of society and the development of economy, the problem of urban traffic congestion has gradually emerged. More and more cities use intelligent transportation systems to regulate traffic flow and optimize the use efficiency of urban road networks. With in-depth research, intelligent transportation systems are gradually moving towards Intelligent, dynamic and informatization. Relevant personnel obtain real-time traffic status data and a large amount of historical traffic data as research objects, and learn the evolution trend of short-term traffic status through calculation.

[0003] Traffic flow forecasting is mainly to realize the calculation of the number of traffic entities passing through a certain point, a certain section or a certain lane of the road...

Examples

Embodiment 1

[0028] Embodiment 1: as figure 1 As shown, a traffic flow prediction method based on sliding window average includes the following steps:

[0029] 1) Collect historical traffic flow data and forecast day traffic flow data;

[0030] 2) Set the window threshold and train the parameters of the traffic flow prediction model; the traffic flow prediction model is:

[0031] X ^ = C · Φ T - - - ( 1 )

[0032] in, Represents the predicted traffic flow data matrix in a continuous interval; C represents the parameter matrix; Φ is the eigenvector matrix describing the changing trend of traffic flow in the corresponding interval;

[0033] The specific calculation process of the parameters of the training traffic flow forecasting model is as follows:

[0034] 2.1) Define the data in M ​​rows and ω colu...

technical field

[0001] The invention relates to the field of intelligent traffic science, in particular to a traffic flow prediction method based on sliding window averaging. Background technique

[0002] With the progress of society and the development of economy, the problem of urban traffic congestion has gradually emerged. More and more cities use intelligent transportation systems to regulate traffic flow and optimize the use efficiency of urban road networks. With in-depth research, intelligent transportation systems are gradually moving towards Intelligent, dynamic and informatization. Relevant personnel obtain real-time traffic status data and a large amount of historical traffic data as research objects, and learn the evolution trend of short-term traffic status through calculation.

[0003] Traffic flow forecasting is mainly to realize the calculation of the number of traffic entities passing through a certain point, a certain section or a certain lane of the road...

Examples

Embodiment 1

[0028] Embodiment 1: as figure 1 As shown, a traffic flow prediction method based on sliding window average includes the following steps:

[0029] 1) Collect historical traffic flow data and forecast day traffic flow data;

[0030] 2) Set the window threshold and train the parameters of the traffic flow prediction model; the traffic flow prediction model is:

[0031] X ^ = C · Φ T - - - ( 1 )

[0032] in, Represents the predicted traffic flow data matrix in a continuous interval; C represents the parameter matrix; Φ is the eigenvector matrix describing the changing trend of traffic flow in the corresponding interval;

[0033] The specific calculation process of the parameters of the training traffic flow forecasting model is as follows:

[0034] 2.1) Define the data in M ​​rows and ω colu...

Images