Research On Performance Uncertainty Analysis Methods For Distributed Satellite System In Conceptual Design Phase

In space science missions,distributed satellite systems can form large-scale observation baselines in space or use spatial configurations and orbital motions to form targeted multi-sampling arrays.The signals can be effectively improved by correlation processing of multi-point sampling.Thus,the overall performance of space science exploration and earth observation missions can be enhanced.Through this detection method,distributed satellite systems can observe more physical phenomena and astronomical data,making spatial exploration further developed.For space science missions,the detection process of a satellite system can be thought as a measurement by recording a time series of the detection values.These measurement results not only include the physical quantity of the payload record,but also the physical parameters of the satellite platform,such as position,attitude,time,temperature,etc.In preliminary design phase of a satellite system,the specific parameters of the satellite platform,orbit,and payload need to be designed.Furthermore,uncertain parameters such as satellite positioning accuracy,attitude measurement accuracy,clock accuracy,and payload measurement accuracy also need be consideration.The characteristics of these uncertainty parameters are time-varying errors of the measurement results in the detection process of satellite system.In the past,quantitative analysis between these uncertainty parameters and the accuracy of the final detection result usually uses a statistical method based on sampling simulation.The time efficiency is low,and the importance of each uncertainty parameter cannot be evaluated.Thus,the main research work of this paper is as follows:1.The expression of uncertainty parameters in distributed satellite systems and the analysis of error transfer process in the detection are studied;Based on the traditional variance-based analysis methods,the time-varying errors in distributed satellite systems are analyzed.The role of uncertainty parameters in the system,and the specific parameters are used as an example to analyze the error transfer process of uncertainty parameters.2.In order to solve the issues both caused by the time consuming and the lack of explicit expression of the different parameter importance in the traditional uncertainty analysis which using the sampling-based simulation method, combined with the actual needs of uncertainty analysis in the conceptual design phase,the method of using normal distribution as a model for describing the uncertainty of complex systems.This method can reduce the dimension of a high-dimensional nonlinear system into a linear combination is proposed.The influence coefficient of different uncertainty parameters are obtained by constructing a weight matrix.Thus,to enhance the time efficiency in uncertainty analysis,meanwhile,to apply analysis to the importance of different parameters and lay the foundation to support the uncertainty parameter analysis in the conceptual design of distributed satellite system.3.An artificial neural network method is proposed as a substitute model for uncertainty analysis method.Using the nonlinear properties in fitting regression analysis by artificial neural networks,a specific neural network structure is designed.After experimental verification,a small amount of simulation results can be used as a training sample to achieve convergence,and effectively represent the original properties of the replaced system.This paper is focus on the study of the analysis method of uncertainty parameters causing time-varying errors in the design stage of distributed satellite systems.Through the simulation model of the data inversion algorithm of the distributed satellite system,a weight-based regression analysis method and two artificial neural network-based analysis methods are introduced based on the traditional sampling simulation method.Simulation experiments and verification analysis are carried out using two typical distributed satellite systems.By comparing with the traditional sampling-based simulation statistical methods,the effectiveness and time efficiency of the method are verified.It provides method support for the analysis of uncertainty parameters in the design stage of distributed satellite systems.