Font Size: a A A

Research Of Actuator Fault Diagnosis For Underwater Vehicle Controlled By Rudders And Thrusters

Posted on:2012-12-12Degree:MasterType:Thesis
Country:ChinaCandidate:N GaiFull Text:PDF
GTID:2218330368983050Subject:Mechanical and electrical engineering
Abstract/Summary:PDF Full Text Request
The ocean as a source of human survival and development has an increasingly important position, with the declining of the non-renewable resources on land. The exploration of the marine resources has drawn more and more attentions. The underwater vehicle as the only equipment that could be able to work in the deep sea environment has an irreplaceable role in ocean development. The autonomous underwater vehicle (AUV) works in the ocean environment without man and cable, and completes tasks depending on its intelligent decision-making system independently. The security is a key problem in the process of AUV researching and practicality. Studies of the fault diagnosis technology, which is a basic technology of AUV security, have great research significance and practical value.Researches have been conducted on the establishment of dynamic model for underwater vehicle, fault diagnosis and isolation of rudder and propeller system, aiming at the actuator fault diagnosis problem of the rudder and propeller joint control underwater vehicle.Taking the rudder and propeller joint control underwater test prototype named "HaiYi-I " as the research object, a model establish method, which combines the analytical model and neural network together, is proposed, in order to improve the accuracy of the dynamic model. Deviation of surge velocity and heading angular velocity output between the underwater vehicle dynamic analytical model and the actual underwater vehicle are fitted through the improved Elman neural network model. Output of analytical model is modified according to the output deviation of neural network model. Modeling error caused by model simplification and outside interference of analytical dynamic model is decreased. Results of simulation and water-tank experiments shows that the dynamic establishment method of "HaiYi-I "rudder and propeller joint control underwater vehicle proposed in this paper, has a higher accuracy than that of the single dynamic analytical model and that of single dynamic neural network model.An actuator fault diagnosis method based on the inverse system is proposed, aiming at the actuator faults diagnosis problem of the rudder and propeller joint control underwater vehicle. Firstly, analysis of reversibility has been conducted to ensure the reversibility of the rudder and propeller joint control underwater vehicle and its actuators. Then, establish the inverse model of all the actuators by the improved Elman neural network. Control variable corresponding to the actual state of actuators is obtained based on the inverse model of actuators. Comparing this control variable and control variables of the underwater vehicle controller output generates the control-variables residuals, which could reflect the faults conditions of actuators, and be used to diagnoses the time-varying actuator faults. Results of test prototype simulation and water-tank experiments verify effectiveness of this method in fault diagnosis of actuator system.An actuator fault isolate method based on the inverse system is proposed, aiming at the problem that the symptom of rudder and propeller joint control underwater vehicle when fault of rudder and propeller system occurs is so similar that it's hard to isolate faults only by surge velocity residuals and heading angular velocity residuals. Control variable output of the inverse model is obtained through the inverse model, aiming at the faults of single actuator. Outputs of rudder and propeller inverse model, as well as actual control and state variables of AUV in previous time are substituted into dynamic model of the underwater vehicle to solve the state variables at present, thanks to the characteristic that control variables exported by the inverse model of actuator in fault is the most close to the actual state of actuator. The actuator inverse model, of which the output state variables are the most close to the actual state variables, is the inverse actuator of actuator in fault. A fault isolation method based on the inverse system is brought forward, aiming at the isolation problem of actuator simultaneous faults. By adding the predetermined control variables to each actuator, autonomous diagnosis of each actuator is conducted, and simultaneous faults isolation of underwater vehicle actuator is realized. Results of simulation and water-tank experiments validate the effectiveness of this method in isolating the simultaneous faults of actuator.
Keywords/Search Tags:underwater vehicle controlled by rudders and thrusters, fault diagnosis, neural network, inverse model
PDF Full Text Request
Related items