Font Size: a A A

The Application Of Neural Networks And Genetic Algorithms In The Mg / Ptfe-rich Propellant Formulation Optimization

Posted on:2011-11-12Degree:MasterType:Thesis
Country:ChinaCandidate:L QiaoFull Text:PDF
GTID:2192360302498611Subject:Military chemistry and pyrotechnics
Abstract/Summary:PDF Full Text Request
Refer to the composition optimization of Mg/PTFE fuel rich propellant, this thesis presents a method of formula optimization by the neural networks combined with genetic algorithm after analyzing the traditional optimize methods and their shortcoming. Firstly, the algorithm and structure of three different network models——BP,GRNN and SVM were studid on the base of the basic principles of neural networks and genetic algorithm. secondly, the uniform design method was used to design the experimental formula of Mg/PTFE fuel rich propellant,and the combustion heat, combustion temperature and combustion rate were measured for the network trainning.After that,the three network models were used for the modeling and prediction of the experiment data respectively, the results showed that,the prediction error of SVM were less than 10%,so it can be used to solve the problem in this thesis. Then, the genetic algorithm was used to get the best formula with the propellant property predictions as optimization goal, the best formula is:the ratio of Mg to PTFE is 0.491, the binder content is 12.5%,the diameter of Mg is 26.90μm, the diameter of PTFE is 26.90μm.Next,the performance of the best formulas was tested, the results showed that their combustion heat, combustion temperature and combustion rate were all at a high level,and they also had low sensitivity and high stability which can meet the requirements of ramjet,so they are worth further study. Moreover, this thesis also built a visible user interface of the prescription optimization system by GUI which can be used in the study on propellant and other composite pyrotechnic system.
Keywords/Search Tags:Fuel rich propellant, Uniform design, Artificial neural networks, Genetic Algorithm, Composition optimization
PDF Full Text Request
Related items