Travel mode identification method based on big data machine learning

Abstract

The invention discloses a travel mode identification method based on big data machine learning, which comprises the following steps: extracting a training sample investigation object and dispatching an acceleration detection device; Collecting mobile phone signaling data and acceleration detection device data; Analyzing the data fluctuation characteristics, getting the velocity and acceleration fluctuation characteristic data as the forecast input value, travel mode as the output value; extracting 80% of that data from the input parameter set as the input data set, and adopting the corresponding travel mode set as the output set to train the machine learning algorithm, and the prediction accuracy of each algorithm is detected with the remaining 20% of the data; Machine algorithm training achieves more than 80% prediction accuracy, can be used to divide the travel mode, setting the input data set as the velocity acceleration characteristic data set, the algorithm output value is the travel mode. The travel mode identification method of the invention has the characteristics of small workload and high identification accuracy.

Description

technical field [0001] The invention belongs to the field of traffic information engineering and control, in particular to a travel mode identification method based on big data machine learning. Background technique [0002] The prediction of travel mode distribution is an important part of urban road traffic planning. Its purpose is to predict the utilization of various transportation modes by urban residents within the planning period. The travel mode choices of urban traffic participants reflect the proportion of passenger flow carried by different transportation modes at the level of the entire city. Traditional traffic surveys are often obtained by OD surveys, and OD surveys based on questionnaire surveys have the problems of heavy workload and the scope of surveys is limited by work capacity. The OD survey data acquisition period is long, the data analysis is more complicated and the repeatability of the survey results is not high, which brings a lot of inconvenience ...

AI Technical Summary

Problems solved by technology

The traffic volume allocation based on the sharing rate is affected by the distribution of survey data and cannot accurately describe the actual traffic volume on the road

[0005] The current travel mode identification mainly includes the aggregate model and the non-aggregate model. Recently, some scholars have proposed the method of utility function or neural network to divide the traffic mode, but these methods all rely on a large amount of data investigation work, from traffic characteristics, Establish a combined model based on the characteristics of user groups or the composition of urban traffic structure and urban traffic modes; it is only applicable to the analysis of the macroscopic travel mode sharing rate, and it is not applicable to the analysis of individuals as a unit

Images