Helicopter Rotor Fault Diagnosis Based On Gray Neural Network Technology

The rotor system provides lift for helicopters and it is also the manipulation mechanism of helicopters,and plays a decisive role in the safety of helicopters and pilots.Helicopter health and usage monitoring system(HUMS)is an important technology in helicopter health management,which effectively improves the safety,reliability,maintainability and economy of helicopters.At present,our country has begun to attach importance to this technology,and some helicopters have installed the system.Helicopter fault diagnosis technology is deficient in the diagnosis of rotor faults,and the establishment of an efficient rotor fault diagnosis model will be of great significance for the domestic helicopter HUMS system.It is difficult to monitor the rotor system directly.It is a good solution to this problem by using the vibration signal of the fuselage to diagnose rotor faults.In this paper,starting with the helicopter fuselage vibration signal,a gray neural network model is established by analyzing gray theory and neural network,and a helicopter rotor unbalance fault diagnosis model is constructed.The main contents of this paper are as follows:(1)Through the analysis of gray theory and neural network,a gray neural network is constructed,which is a combination of convolution neural network and gray relational analysis.The convolution neural network is used to combine and extract the deep features.The gray relational model is used as the classifier to classify,and the convolution neural network is experimentally analyzed to determine its performance.(2)Through the analysis of the basic Denoising Auto-encoder,a Convolution Auto-encoder deep neural network is proposed for signal feature extraction.Add random noise to the input to improve the robustness of the extracted features.At the same time,the traditional signal processing method wavelet is used to process the signal.Compared with the classification results of the two methods,both the classification accuracy are good.(3)The features extracted respectively by Convolution Auto-encoder and wavelet are used in the classification of the tandem gray network model constructed in this paper.The experimental results show that the proposed diagnosis model can effectively classify helicopter rotor faults.