Method for traffic flow prediction based on spatio-temporal correlation mining

摘要

The disclosure includes a method for traffic flow prediction based on data mining on spatio-temporal correlations. The method includes establishing a prediction model, data mining on spatio-temporal correlations, and traffic flow prediction based on spatio-temporal correlated data. The prediction model can be a linear regression model with multiple variables. The data mining on spatio-temporal correlations is based on a multi-factor linear regression model and by means of the optimization method in terms of sparse representation. The data from the spatio-temporal correlated sensors that are relevant to the prediction task are determined automatically. The traffic flow prediction based on spatio-temporal correlated data refers to that the prediction is performed with the input to the prediction model to be the data from the spatio-temporal correlated sensors.

主权项

1. A method for traffic flow prediction based on data mining on spatio-temporal correlations, comprising:
(a) collecting raw data of traffic flows through sensors distributed at nodes located along a road network; (b) preprocessing the collected raw data into a valid form of traffic flow data; (c) establishing a prediction model, comprising: letting νij represent traffic volume data sampled at sensor j at time i; supposing that there are in total m sensors in a road network; denoting the state of the whole road network at time i as Vi=[νi1, νi2 , . . . Vim]; and using a linear regression model to predict the traffic volume data collected at senor j with time lag τ as follows:
νi+τj=Viwj wherein weights wj=[w1j, w2j, . . . wkj, . . . wmj]T are parameters to be optimized and νi+τj is the predicted traffic volume;
(d) mining spatio-temporal correlations, comprising: applying a sparse representation as an optimization method to obtain the parameters wj, wherein wj=[w1j, w2j, . . . wkj, . . . wmj]T represent the spatio-temporal correlations between the traffic flow data from each sensor in the whole road network and the data from the target sensor j undergoing prediction; wherein when wkj=0, the data from sensor k are not correlated to the data from sensor j; wherein otherwise, wkj indicates the correlation degree between the data from sensor k and the data from sensor j, k=1,2, . . . ,m; and (e) performing traffic flow prediction by applying the spatio-temporal correlated data as the input to the prediction model.