| Fault diagnosis of helicopter rotor is an important component of HUMS(helicopter health and usage monitoring system).Measuring rotor hub loads and tip displacement for fault diagnosis is complex and difficult to use in practice.Combined with fuzzy reasoning technology and the principle of neural network, a fuzzy RBF neural network model of rotor fault diagnosis has been proposed in this thesis based on the theory that rotor imbalance fault can been diagnosed by the fuselage vibration signal.The thesis mainly consists of the follows:(1)Principal component analysis and independent component analysis has been used respectively to extract feature of rotor fault signal. Combined with the diagnosis effect of fuzzy RBF neural network model to analyze two kinds of feature extraction methods, the results show that method fault classification accuracy rate of the principal component analysis is 97.06%, while independent component analysis is only 47.05%.(2)The fuzzy RBF neural network model for classification of the rotor imbalance fault diagnosis has been established.In addition,the influence of different input partition, fuzzy subspace and the main contribution rate and other factors on the classification results has been Considered.Compared to the diagnosis results of support vector machine, the neural network diagnosis model of RBF, the results shows that the fuzzy RBF neural network recognition rate is far higher than that of the diagnosis model of support vector machine and RBF neural network.(3)Genetic algorithm is used to optimize the clustering center of fuzzy RBF neural network of the fault degree identify.Compared to relative error of fault degree recognition before and after optimization, the results showed that the optimized recognition accuracy has been improved greatly.(4)Rotor fault diagnosis system interface, which contains 5 function modules, developed by using the GUI module of MATLAB, operate convenient, and can realize fault diagnosis quickly. |
PDF Full Text Request