| Aircraft ground hydraulic oil pump vehicle is considered as an important safeguard-equipment for aviation. As the source of power, hydraulic power is provided when aircraft's engine doesn't work, and the hydraulic oil with corresponding pressure and flow is suppied to the aircraft, so the aircraft's hydraulic systems are inspected whether the work is reliable, such as the aircraft's control system, undercarriage system, etc. By using this equipment, the effective life-span of aircraft's engine can be enhanced, and the hydraulic system can also be purified and washed.A new aircraft ground hydraulic oil pump vehicle and its hydraulic system are introduced, and Virtual Intrument (VI) technology is used to realize the function of data collection and condition monitoring. An advanced fault diagnosis method is elaborated, which is applied to the hydraulic system's fault diagnosis for aircraft ground hydraulic oil pump vehicle.This paper researches mainly several aspects as the followings:(1) According to the oil pump vehicle hydraulic system's technical requirements and the operating features, the working loop and depurative loop of aircraft ground hydraulic oil pump vehicle's hydraulic system are designed, and their principles are also elaborated. At the same time, the choice procedure and performance parameter of frequency conversion motor, transducer, and many other hydraulic elements such as hydraulic pump are introduced. The hydraulic system with different load is simulated by hydraulic simulation software, the working process of hydraulic system is simulated, and the simulation curves are generated. It is concluded that the hydraulic system meets the requirements of specifications and the operating features.(2) The signal collection hardware system's program of condition monitoring system is drawn up using Virtual Instrument (VI) technology, and the interface of visual condition monitoring system is developed by Lab Windows/CVI software. The view-datas of condition monitoring system includes real-time data curves and historical data curves. As a rule, the real-time curves are the intuitionistic description of the current working condition, and the historical data curves can be used to study the former work condition and variational tendency of the oil pump vehicle's hydraulic system. At the same time, the condition monitoring system has also provided the effective foundation for the fault diagnosis technology.(3) It is proved that the data collection and condition monitoring system can work effectively, through carrying out the test to the hydraulic oil pump vehicle. The user interface of condition monitoring system is friendly and easy to operate. The test pressure curves of hydraulic pump outlet are tallied with the pressure changing rules of its load condition. What is more, it is proved that the design of the hydraulic system is reasonable.(4) The BP neural network expert system used in fault diagnosis of hydraulic system is built. According to the fault modes that have been analyzed in chapter 4 and the oil pump vehicle monitoring system's signal contents, we select 10 kinds of common fault phenomenons and 10 kinds of common fault origins. Based on the table of"origin——symptom"which is established by the experience estimation method,the study signals for neural network and the expected output are determined. A feasible choice method on how to choice the node number of hidden layer is proposed. According to the empirical formula of the hidden node number, a neural network whose node number of hidden layer is changeable between 6 to15 is built. Through contrasting the errors of different neural networks, it is concluded that the best number of hidden layer is 11. At the same time, the neural network is trained using the steepest descent BP algorithm and the LM algorithm separately. It is obviously concluded that the LM algorithm has a quicker astringency by contrasting the convergent steps of the two algorithms. Finally, the codes of BP neural network used in fault diagnosis are compiled. MATLAB neural network toolbox is used to train the samples, and an instance is emulated to prove that BP neural network has great astringency, reliability, and predictability. This provides a new reliable method for aircraft hydraulic oil pump vehicle hydraulic system's fault diagnosis. |
PDF Full Text Request