Research On Traffic Fusion Spatio-temporal Prediction Model And Dynamic Path Planning Algorithm

As an emerging transportation service system,intelligent transportation makes full use of high-tech Internet technologies,for instance,Internet of things and artificial intelligence to ensure the operation and management of urban transportation.Traffic flow prediction and route planning is the core technology of intelligent transportation.It can not only provide reference for people’s travel,but also provide strong technical support for alleviating traffic congestion and traffic accident avoidance.However,the current prediction model ignores the dependence between the functional zones formed by the functional features of the traffic network.In addition,the effect of external factors,for instance,weather and time on the prediction is also ignored.Due to the dynamics and uncertainty of transportation and the complexity of road network,how to consider the impact of future traffic information in route planning algorithm and improve planning efficiency in dynamic road network are also problems to be solved.To address the above problems,this paper investigates the spatio-temporal traffic flow prediction and path planning algorithm,and the main work is as follows:(1)A hierarchical mapping and interactive attention data fusion network for traffic prediction.In terms of extracting spatial features,the model uses hierarchical mapping network to construct functional zones according to traffic features and capture the correlation between areas.In terms of temporal features,the model adopts the inter-attention mechanism to fuse the traffic data and the external factor data,and calculate the impact of external factors on the traffic data.Finally,the gating fusion mechanism is used to further combine the temporal and spatial characteristics to obtain the prediction results.The experiment shows the influence of different types of external factors and different data fusion methods on traffic flow prediction.In addition,compared with the baseline models,the prediction results of the proposed model in complex traffic network are more accurate.(2)A fast dynamic urban route planning algorithm based on future traffic changes.The algorithm uses the predicted average speed to calculate the estimated travel time of each road,and takes the estimated travel time as the dynamic weight of the road.In order to improve the efficiency of route planning in complex road network,this paper introduces Highway Hierarchies algorithm to preprocess the traffic network.The algorithm reduces redundant route search by extracting critical route and constructing a traffic hierarchies.For the hierarchical road network structure,this paper uses the optimized bidirectional search strategy to find the optimal route.Experiments verify the influence of Highway Hierarchies algorithm hyperparameter parameter setting on the efficiency of the algorithm,and provide a certain reference for the parameter setting of subsequent experiment.In addition,experiments also prove that the proposed model has higher efficiency and shorter actual travel time than baseline models.