Research On Fault Detection And Diagnosis Of Liquid Propellant Rocket Engine Based On Improved Optimization Algorithm

Liquid rocket engine is a kind of complex power system.With the increasing payload of carrier rocket,the engine works in more and more complex and harsh environment.Once the failure of liquid rocket engine occurs,it will lead to the failure of the space system.Only the engine fault detected in time and inerrantly can the emergency measures be taken to prevent degradation in engine performance or even engine damage.It is of great engineering significance to detect the engine failure in time and accurately for the success of space launch mission.Based on the improved optimization algorithm and using the historical test data of a certain type of liquid rocket engine,this paper transforms the core steps in the research of fault detection and diagnosis of liquid rocket engine into corresponding optimization problems for processing.The main research contents include:(1)Based on the test platform of a certain type of liquid rocket engine,the detailed fault modes are obtained by using the methods such as historical fault statistical analysis and FEMA(Fault Mode and Effects Analysis),which provide the basis for the selection of monitoring parameters of the fault detection and diagnosis model.(2)Based on the QGA(Quantum Genetic Algorithm)in the optimization algorithm,a fault detection of liquid rocket engine method based on QGA to optimize BP(Back Propagation)neural network was implemented.The dynamic improvement strategy and catastrophe strategy are selected as the operation criterion of QGA evolution.The initial weight system of BP neural network is optimized by improved QGA.A fault detection model based on QGA-BP is established for a certain type of liquid rocket engines.The simulation results show that QGABP can effectively overcome the drawbacks of BP neural network easily falling into local optimum and GA(Genetic Algorithm)similar exhaustion.Compared with traditional BP neural network and single GA,QGA-BP has higher convergence speed,evolutionary algebra and fault detection accuracy,and can be effectively used in fault detection of liquid rocket engines.(3)Based on CS(Cuckoo Search)algorithm in the optimization algorithm,a fault detection method of liquid rocket engine based on CS algorithm optimized BP neural network is realized.The CS algorithm is used to optimize the BP neural network.The simulation results show that the fault detection model based on CS-BP neural network has fast convergence speed and high accuracy.It can be used as a supplementary method in the fault detection algorithm of the actual test run,and the test results of QGA-BP model are verified and analyzed by CS-BP model.(4)Based on PSO(Particle Swarm Optimization)in the optimization algorithm,a fault diagnosis algorithm of liquid rocket engine based on CPSO(Chaotic Particle Swarm Optimization)and LSSVM(Least Square Support Vector Machine)is implemented.The algorithm optimizes the regularization parameters and the width of the kernel function ofLSSVM fault classification model based on the global search ability of CPSO.First,a chaotic sequence is used to initialize the particle positions,thereby enhancing search diversity.Then,the premature convergence of normal PSO is countered by using CPSO.If PSO falls into a local optimum,the extreme position of the population is adjusted and the current search trajectory of the particles is disturbed,thereby increasing the probability of escaping the local optimum.The simulation results show that the CPSO-LSSVM classification model has higher classification accuracy than the GA-LSSVM classification model and the LSSVM classification model based on cross validation,and has higher diagnostic efficiency.It is more effective in the application of fault diagnosis for liquid rocket engines.