Enhancing Satellite Intelligent Prediction with Parameter Correlation and LSTM Multidimensional Forecasting

The satellite operates long-term in a complex environment, influenced by various uncertain factors, causing fluctuations in its performance and functionality. Satellite telemetry parameters are crucial indicators for assessing the satellite's status, and predicting these parameters plays a significant role in determining the satellite's condition. In order to improve the accuracy of satellite telemetry parameter predictions, this study proposes a correlation-based LSTM multidimensional forecasting model. By selecting real satellite telemetry datasets and analyzing the correlations among telemetry parameters, the model combines strongly correlated parameters for simultaneous prediction, which reduces errors to some extent. Experimental results demonstrate that for parameters Battery_U1 and Battery_U2, when combined with strongly correlated parameters for prediction, their final Mean Absolute Error (MAE) decreases by 35.85%and 52%, respectively, and their final Mean Squared Error (MSE) decreases by 29.41% and 52.01%, respectively.