Research On Real-time Filtering And Prediction Of The TT&C Ship's Attitude

When the spacecraft tracking,telemetering,and control(TT&C)ship is sailing on the sea,it will sway due to the influence of sea waves,wind,and other environmental factors.For high-precision measurement,it is essential to filter the sway data in real time and calculate the multi-step prediction values of sway angle,angular velocity,and angular acceleration.At the same time,predicting the attitude of the ship a few seconds or more in advance can further improve the control accuracy of the anti-rolling device,reduce the ship's sway,and improve the ship's seakeeping.Thus,studying the data characteristics of the ship's attitude and its filtering and forecasting methods have a strong significance for improving the comprehensive measurement accuracy at sea.The main jobs are stated as follows:(1)Based on the measured data,the characteristics of the sea attitude data of the TT&C ship are analyzed.Firstly,the spectrum characteristics analysis shows that the primary signal of the ship shake data is distributed in the low-frequency range of 0.6 Hz,and the noise characteristic analysis indicates that the noise is non-Gaussian white noise.Secondly,the high-order analysis of ship shake data shows that the jounce periodicity is significantly weakened after four differentials,and the attitude data can be regarded as the jounce noise-driven process in a short time.At the same time,the chaotic characteristics of sway data are analyzed qualitatively and quantitatively,and the delay time and embedding dimension of roll and pitch data are determined.(2)Based on the chaotic characteristics of ship rock data,the maneuvering target tracking theory is introduced to study ship sway filtering and multi-step forecasting.Aiming at the problem that the existing model order is not high enough,a higher order model is derived and a "current"statistical-Jounce(CS-Jounce)model adaptive method based on maximum correntropy is proposed.Simulation results show that the proposed method can estimate and predict the ship's angular velocity and angular acceleration well.It has high filtering and forecasting accuracy and reliable robustness.(3)On the basis of data filtering,the extremely short-term forecast of ship shake data is further studied.Based on the analysis of ship sway data characteristics,the data structure and phase space reconstruction are used to simplify the model structure.The efficiency and accuracy of offline training and online forecasting method using radial basis function(RBF)neural network are analyzed.It is pointed out to be suitable for online prediction of roll and pitch data.The kernel recursive least squares(KRLS)algorithm based on phase space reconstruction can be used for online prediction of roll and pitch data.This method can achieve higher prediction accuracy than RBF network and has higher computational efficiency.For the non-stationary heading data,the method of Hodrick-Prescott(H-P)trend extraction combined with RBF network forecasting is proposed.The simulation results show that the proposed method can effectively realize the online forecast of heading data and overcome the accuracy of the simple RBF network method in non-stationary data forecasting.problem.