Research Based On BP-ANN And Software Development On The Shock Environment Of Satellite-rocket Separation

Satellite-rocket separation is the last link of rocket launching mission, determined the missionwill be success or failure. Usually, satellite and rocket are connected by the clamp band andseparated with explosion of the explosive bolts. During the separate process, because of therelease of strain energy of preload, there will be a high frequency shock environment acting on thesatellite.Considering the shock environment has an important affection to the satellite, this thesisaimed to research on this problem, proposed a solution based on Back-Propagation neural networkand some kinds of its modified forms, seeking for the mapping relationship between theparameters and the maximum shock response of the satellite, and using the relationship to predictthe maximum shock response of other satellites.In order to do this, the thesis, first of all, researched on Shock Response Spectrum (SRS), andcalculated the SRS and maximum shock response of the sample satellites. Then this thesisestablished BP net works and several kinds of its modified forms (such as Additional momentummethod, Variable learning rate method, LM method, GA-BP method) to seeking for the mappingrelationship between the parameters and the maximum shock response of the satellite, and usingthe relationship to predict the maximum shock response of other satellites. Then, a satellite-rocketseparation shock environment calculating software will be developed based on the MATLABGUIDE platform. The software will use the theory analyzed in the front part of this thesis and willhave the functions of data management, establishing network, predicting shock response, savingresult and so on. In the last part of this thesis, a example will be given to show that the theory andsoftware this thesis proposed is right.