Research on Application of GA-BP Artificial Neural Network Model for Launch Centre Temperature Forecast
In order to reduce the temperature forecast error in the launch site,the GA and BP were combined.Based on the EC fine grid forecast data and the launch site observational data from 2018 to 2022,the prediction factors were screened using the correlation coefficient.In the end,the tem-perature forecast model of the launch site was established.The results showed that the mean absolute error of the model temperature forecast was 1.132℃which was 7.8%better than that of the EC fine grid;The error standard deviation of the model temperature forecast was 0.907℃which was more stable than that of the EC fine grid and the manual correction work was greatly reduced;During the support process of Shenzhou-15 manned space mission,the window temperature predicted by the model was-18.08℃,and the actual temperature was-17.9℃.It could provide higher reliable forecasting method for temperature precision forecasting,especially in critical conditions.
BP neural networkgenetic algorithmspace launch supporttemperature forecast