A Distributed Traffic Flow Prediction Method and System

Abstract

The invention discloses a distributed traffic flow prediction method and system. The method includes: step 1, obtaining traffic flow data; step 2, using a pre-established distributed traffic flow prediction model to process the traffic flow data to obtain prediction Result; Wherein, the generating method of described distributed traffic flow forecasting model is: step 2.1, calculates based on traffic flow data and measures the single road section congestion value; Step 2.2, trains single road section traffic flow prediction model based on described actual measured single road section congestion value; Step 2.3, using the deep random forest, the traffic flow data of the selected area and the predicted single-segment congestion value output by the single-segment traffic flow prediction model are input as training set data to train the distributed traffic flow prediction model. In the present invention, calculation is shared by each node to perform parallel calculation, which greatly reduces the calculation amount and calculation time of the server.

Description

technical field [0001] The invention relates to the field of real-time traffic, in particular to a distributed traffic flow prediction method and system. Background technique [0002] Urban traffic flow prediction is an important component of smart cities. Traditional traffic flow forecasting technology, the model is simple and easy to use, but it is also limited by the algorithm, which cannot well reflect the complex and changeable characteristics of traffic flow. In recent years, algorithms led by neural networks have shown higher accuracy in traffic flow prediction, but we also need to see their inevitable shortcomings: [0003] 1. Based on the neural network algorithm and its improved model, it takes a lot of time and uses a lot of manually marked data during the construction process. After the construction is completed, it can be deployed to the corresponding monitoring equipment. [0004] 2. As time changes and cities develop, the existing algorithm models may not be...

AI Technical Summary

Problems solved by technology

[0004] 2. With time changes and urban development, the existing algorithm models may not necessarily meet the latest prediction needs. At this time, it is necessary to mark, train, and deploy according to the latest data, which consumes a lot of manpower and material resources

[0005] 3. The trained model is a whole. If part of it is changed, the whole model needs to be replaced; all calculations are placed on the server where the model is stored, which makes the server have greater computing pressure

Images