Wuhan meteorological observation elements based on chaos optimization BP neural network quality control method

In this paper, based on the changing characteristics of the same observation element with time, analyzes the chaotic characteristics of the time series of several common meteorological observation elements of vectors and non-vectors firstly, and then reconstructs the phase space of the time series of meteorological elements after inserting errors according to the chaotic characteristics of each element. The time series of meteorological elements reconstructed by each of Wuhan station's 800 hourly data of temperature, pressure, wind direction and precipitation in winter and flood season in 2016 were taken as the input of BP neural network, and the corresponding quality control results were taken as the output, and establishes a BP neural network meteorological data quality control model based on chaos optimization. Cross validation is used to train the model and improve the precision of model quality control. The results show that the difference between the estimated value and the actual value of the model is small, and the missing value can be fitted and the outlier can be detected. It has a good guiding significance for the quality control of Wuhan surface meteorological observation data.