Liquid Rocket Engine Fault Diagnosis Based on KPCA-ICOA-BP Model
In order to improve the reliability of the liquid rocket engine,aiming at solving the liquid rocket engine fault diagnosis problem,a kind of fault diagnosis model based on the kernel principal component a-nalysis(KPCA)and the ICOA-BP algorithm is proposed in this paper.The feature extraction and dimen-sionality degradation of measured parameters are implemented by using KPCA algorithm which ensures suf-ficient features amount extracted and coming together by reducing the complexities of the data and the com-putational cost.And an improved Coati optimization algorithm(ICOA)is proposed to optimize the BP neu-ral network,aiming at improving the diagnostic accuracy of the BP neural network.The algorithm is vali-dated by using liquid-oxygen-methane rocket engine test data,and the experimental results show that the ICOA-BP algorithm exhibits faster convergence speed and higher optimization finding accuracy compared to the COA-BP algorithm.The diagnostic accuracy of ICOA-BP algorithm can reach 96.5％on the data ex-tracted from KPCA features,which is respectively 3.5％and 3％higher than the diagnostic accuracy of BP neural network and Support Vector Machine(SVM).Compared with particle swarm algorithm(PSO)and genetic algorithm(GA),ICOA-BP algorithm demonstrates better searching ability for the global optimal solution.
NETL
NSTL
万方数据
