The invention discloses a highway pavement
temperature forecasting method based on physical constraints, which comprises the following steps of: collecting and preprocessing meteorological data of all stations in a region, historical
observation data of pavement temperature and geographic data, and adding
time information to construct time features; building a U-net neural
network model, carrying out space
field data correction on the multi-
station multi-meteorological element forecast data, and outputting corrected meteorological data of each
station; constructing a bidirectional LSTM
network on the basis of the U-net model,
processing time sequence data, and evaluating and optimizing model performance by using a
loss function; physical constraints including the time relation of temperature change, seasonal and day and night change characteristics, material thermal characteristics, a
thermal balance equation and the like are added into the model, and a model result is optimized; and evaluating the generalization ability and forecasting precision of the model by using a
test set, and finely adjusting
model parameters. According to the method, the precision and reliability of
road surface temperature prediction are remarkably improved, and a reliable basis is provided for
traffic management and safety early warning.