Bus load similar day extraction method based on spatial-temporal characteristics and vector database

The invention discloses a bus load similar day extraction method based on spatial-temporal characteristics and a vector database, relates to the technical field of bus load prediction, and solves the problem that influences of external factors such as rainfall, wind speed, irradiation and cloud cover data are not considered in similar day selection in the prior art. And the selected similar date cannot be ensured to be the date with the highest similarity. According to the method, in the similar day selection process, the spatial-temporal characteristics of the bus load are considered, and the spatial-temporal characteristics of external factors such as air temperature, rainfall, wind speed, irradiation and cloud cover data are also considered. Besides, based on the spatio-temporal characteristics, correlation extraction is carried out by combining time sequence characteristics such as a mean value, a median, a variance, skewness and kurtosis of load time sequence data, and finally, based on cosine similarity