| With the intensifying energy crisis and greenhouse effects,renewable energies are becoming more and more important and attractive.Among the different forms of renewable energies,the tidal current energy has received increasing attentions for academia and industry in recent years due to its high predictability,large reserves and lower dependences on weather changes.Affected by the harsh marine environment,a tidal current energy-based turbine system is prone to faults in its power conversion chain,especially for the parts of generator and power converters.Hence,to improve reliability and availability of the turbine system,a 5-phase non-sinusoidal back electromotive force permanent magnet synchronous generator(5-phase non-sinusoidal back-EMF PMSG)is utilized in this dissertation.The main merit of the five-phase PMSG over conventional three-phase ones lies to its continuous operations under one-or two-phase losses.This dissertation thereby concentrates on open phase faults in the five-phase PMSG and open switch faults in its directly connected machine side converter.Although the five-phase PMSG is able to provide the higher fault-tolerance than the classical three-phase machines,the torque ripples and power perturbations are still high under the faulty conditions,which degrades the tidal energy harnessing efficiency.To address these issues,developing efficient fault diagnosis(fault detection and localization)and fault-tolerant control techniques becomes thus two key factors.The main works and efforts of this dissertation are summarized by following 5 aspects corresponding to the chapters:In chapter 1,research background and significance are firstly illustrated relying on the interests of the tidal current energy.Then the tidal current based turbine projects within and outside China are then collated to observe the gaps and shortcomings.Architectures of different tidal current turbine systems are introduced to select a topology that is equipped with the direct-driven five-phase PMSG and back-to-back converters.The tidal current turbine systems are highly vulnerable to harsh ocean conditions and the high fault rates mainly exist in the generators and converters.In order to increase the systematic availability,5-phase non-sinusoidal back-EMF PMSG mentioned earlier is chosen,whose advantages are comparatively analyzed in comparison with 5-phase sinusoidal and 3 phase sinusoidal back-EMF PMSGs.After that,machine side open phase and open switch faults are chosen as the studied fault types in order to further enhance the fault tolerance of the five-phase PMSG.The existing main categories of fault diagnosis and fault-tolerant techniques,as well as the related works around five-phase PMSG applications are overviewed.Finally,main objectives,work content and dissertation structure are summarized.In Chapter 2,all main parts of the power conversion chain of the studied tidal current turbine system are modeled.The dynamic model of the PMSG is built taking into account the decomposed principal and the secondary sub-machines respectively under orginal,Concordia’s and Park’s frames.Subsequently,basic control strategies for the machine and grid sides are designed on the basis of the constructed model,aiming at controlled operations in healthy conditions.In addition,optimal control references for the torque/current controllers were designed to minimize copper losses.After this,a megawatt-scale back-toback simulation setup is built in MATLAB/SIMULINK.In this,the performances of different simulation modelling approaches are compared and the final choice is the equivalent circuit approach to construct the 5-phase non-sinusoidal back-EMF PMSG.This simulation setup is used to verify the system performance under constant and variable tidal current speed operational modes.In Chapter 3,it investigates fault detection and localization issues around the studied system.To handle the lengthy detection time of conventional methods based on moving average calculation,a new method based on the singularity detection using the stator current waveforms is considered in this chapter.Furthermore,this method does not require the specific system model and might be adapted to other related applications.The proposed method is designed via the abrupt change detection theory that adopts the way of a new filter derivative algorithm.This is realized by a second-order generalized integrator with fasttracking dynamic behaviors.The parameter tuning of the proposed filter derivative algorithm is illustrated in Laplace and temporal domains.Considering the sensitivity against noises of the proposed algorithm,a fault detection and localization strategy is then proposed through an indicator enhancement mechanism,an adaptive detection law as well as a stable fault localization method.The indicator enhancement mechanism relies on the instantaneous responses reflected in the phase current waveforms at the faulty moment,which is achieved by weight assignments to highlight the useful features among the fault indicators.In addition,an adaptive threshold is developed depending on a root mean square value of the five-phase current,to increase the applicability of the strategy under different working conditions.By simulation verifications,it can be found that this proposed fault detection and localization strategy facilitates real-time implementations.This proposed strategy does not require the information of system model and it can effectively reduce the false alarms and missing alarms of fault detection and localization.In Chapter 4,developing an advanced active fault-tolerant control strategy is the main concern,to maintain the operational availability of the five-phase PMSG tidal current turbine system in post-fault modes.Most of the existing literature is based on switching the corresponding fault-tolerant control reference signals under various faulty conditions.However,this needs precise system models with respect to each kind of fault,which may result in heavily computational burden.To overcome this drawback,an active FTC strategy without prior knowledge of fault categories and faulty models is then investigated based on disturbance observer theories.This proposed strategy is realized by a powerful disturbance observer,the generalized proportional integral observer(GPIO).The GPIO is able to achieve the functions of both the fault detection and fault-tolerant control in the same time via an estimation and compensation mechanism.Parameter tuning issues of the estimation and compensation are then analyzed by means of the estimation bandwidths and minimized degrees of torque ripples in the event of faults,respectively.The feasibility of the proposed active FTC is verified through simulation tests using various compensation parameters.This proposed strategy is also compared with other classical methods based on switching different fault-tolerant reference signals(e.g.,equivalent current amplitude and minimum copper loss fault-tolerant control).From the obtained simulation results,it is found that the proposed method achieves an equivalent reduction in torque ripple without the need to know the fault categories and faulty system models in advance.In Chapter 5,a small-scale experimental platform is firstly set up based on dSPACE and a 3.3 kW five-phase PMSG.In this platform,the tidal current turbine is emulated by using a speed-adjustable DC motor.This experimental platform,implemented in the IREENA laboratory,is used to validate the proposed fault detection and localization and the proposed active fault-tolerant control strategies.From the acquired experimental results,it is verified that proposed fault detection and localization strategy(developed in Chapter 3)is highly robust to torque and speed variations.Compared with two classical strategies,moving average based and Luenberger observer-based ones,the proposed strategy is capable of detecting and localizing the single and multiple open-circuit faults more quickly and accurately.Furthermore,the performance of the proposed active FTC strategy(developed in Chapter 4)is verified by evaluating its performance under different faulty conditions.The experimental results doubly confirm that this proposed active FTC strategy can effectively reduce torque ripples without the prior knowledge of the fault types and system model under faulty conditions.In the end,a summary of the research work that has been completed is presented.Some of the outstanding issues and challenges will be addressed as future tasks. |
PDF Full Text Request