Low visibility weather is one of the main factors restricting traffic travel,so forecasting low visibility in advance can help make decisions in advance,avoid risks and reduce losses.Since low visibility weather is a time series issues,the long and short-term memory network of deep learning can significantly improve the forecast accuracy in recent years, multivariate data is used to predict the low visibility of Maotai Airport.Correlation analysis,using Pearson similarity to screen for factors with high correlation,thereby reducing the complexity of the data.The LSTM network is then used to model multivariate time series to predict airport low visibility.After experiments,the model’s accuate forecast rate of low visibility at Maotai Airport was 85.43%,which provided a new forecasting method for airport for low visibility.