Weather prediction model based on CNN-BiLSTM-Attention

Weather data forecasting is one of the hot research fields in time series analysis. The most classic research is to establish a weather sequence analysis model based on the assumption of stationarity. However, this method ignores the correlation between multiple meteorological factors, and cannot truly achieve accurate prediction of multiple factors in the weather sequence. Based on weather time series data, this paper analyzes the current research status of temperature, wind speed and precipitation forecasting at home and abroad, as well as the research and comparison of various time series forecasting models. Based on the above analysis result, weI construct a CNN-BiLSTM-Attention weather forecasting model, and conducts forecasting on the meteorological data of Shaoguan area, and evaluates the performance of the forecasting model. The experiments prove that the CNN-BiLSTM-Attention model has further improved the accuracy of weather forecasting.