Running track prediction method aiming at specific vehicle potential group

Abstract

The invention relates to a running track prediction method aiming at a specific vehicle potential group. The running track prediction method comprises the following steps: looking up the potential group vehicle of a certain specific vehicle by using the original traffic data when the specific vehicle is discovered; judging whether the potential group vehicle is the specific vehicle; increasing the risk coefficient of the vehicle when the potential group vehicle is not the specific vehicle; adding a license number of the vehicle into a specific vehicle list when the risk coefficient of the potential group vehicle exceeds a predetermined threshold value; predicting a running track of the vehicle if the license number of the vehicle is added into the specific vehicle list; performing mode extraction on a running record of the specific vehicle by using a class Apriori algorithm to generate a rule set R during track detection; judging whether a final passing barrier of the specific vehicle exists in the generated rule set R; and looking up a prediction path of the specific vehicle according to the corresponding rule in R if the barrier exists in the rule set R, and otherwise, predicting the running track of the specific vehicle by establishing a Bayesian network of the final passing barrier of the specific vehicle. The prediction result provided by the invention can provide effective technical support for the decision making of related departments and the safety guarantee of urban roads.

Description

technical field [0001] The invention relates to the field of data mining. Aiming at a large amount of original traffic data generated by an urban intelligent traffic information system, a driving track prediction method for a specific potential group of vehicles is proposed. Background technique [0002] Time series mining has become a hot topic in the field of data mining in recent years. Since time series is a new data type with a time relationship between data, it has the following characteristics compared with other types of data: obvious chronological order, can reflect Sequence characteristics, randomness and noise etc. Therefore, in the process of data mining for time series, the time relationship between data in the data set must be considered. At present, the research on time series mining mainly focuses on the following six aspects: time series similarity search, time series classification, time series clustering, time series segmentation and pattern discovery, ti...

AI Technical Summary

Problems solved by technology

However, in the face of such large-scale time-series traffic data, it will take a lot of manpower and material resources to analyze and predict it using traditional statistical methods.

Images