Regional traffic demand prediction method based on convolutional long-term and short-term memory network

Abstract

The invention relates to a regional traffic demand prediction method based on a convolutional long-term and short-term memory network, and belongs to the field of computer data analysis. The inventionprovides a framework integrating multiple tasks. For a traffic flow prediction problem, a multi-task learning layer is added. Through underlying parameter sharing, multiple prediction targets are completed at the same time, and the generalization ability is improved. According to a traditional traffic demand prediction method, only a single prediction task is processed, and the characteristic that multiple related tasks can be parallel is not considered. The idea of multi-task learning is introduced, and multiple tasks are integrated on the topmost layer of the model and trained at the same time, and the situation that each task is trained separately is replaced. By adding the multi-task layer, the shared weight in the network can be fully utilized, and a better prediction result is provided.

Description

technical field [0001] The invention relates to the field of computer data analysis, in particular to a method for obtaining specific time and space information based on deep learning. Background technique [0002] Regional traffic demand forecasting is very important for urban vehicle management. It can help the traffic management platform to better complete the space scheduling of vehicles and reduce the waiting time of passengers. The prediction of regional traffic demand needs to consider factors such as time, space, weather and holidays, so it is challenging. Regional traffic demand is generally based on historical time series data, and mathematical algorithms are used to predict specific location traffic demand, such as time series entropy calculation, Markov chain simulation, distribution balance, Poisson distribution and other algorithms. Deep learning methods have also been widely used in the field of regional traffic demand forecasting, such as using long-term sho...

AI Technical Summary

Problems solved by technology

[0011] In order to solve the technical problems of low prediction efficiency of traditional traffic demand forecasting methods and insufficient use of time-space correlation and external influence factors, the present invention designs a traffic demand forecasting method based on Conv-LSTM network architecture, which can extract the time of traffic demand data Spatial dependence, integrating the impact of external factors such as weather on traffic demand, and integrating multiple tasks to complete the traffic demand prediction of urban target areas

Images