Application Research On Fault Diagnosis Of HVDC System Based On State Estimation Methods For Nonlinear Systems

High Voltage Direct Current (HVDC) is mainly applied to faraway and high capacity power transmission, power system networking. System faults will result in production pause or confusion, so fault diagnosis is the core technology of HVDC and also the precondition of system protection. HVDC system is a nonlinear system, but so far, researches on fault diagnosis of nonlinear systems are few because of their complexity. Also there are a bit of relative studies on faulty diagnosis in HVDC systems inland and abroad, which are still in simulative and experimental stage, and there is a long way to go to put them into use. So we investigate and discuss two state estimation based fault diagnosis methods for nonlinear systems, and put them into HVDC system, the main studying work is shown as follows:Firstly, the basic configuration, principium and characteristic of HVDC system are introduced based on the typical HVDC model present by CIGRE. The platform of EMTDC/PSCAD has been adopted to construct the model, analyze the typical faults of HVDC, and make the simulations.Secondly, a new fault diagnosis method for nonlinear stochastic system with non-Gaussian noise and disturbances based on particle filter is proposed by combining particle filter algorithm and fault diagnosis theory. One of the appealing advantages of the new approach is thai the complete probability distribution information of the state estimates from particle filter is utilized for fault detection, another is its applicability to general non-linear system with non-Gaussian noise and disturbances. The effectiveness of this new method is demonstrated through Monte Carlo simulations and the detection performance is compared with that using the extended Kalman filter on a non-linearsystem.Thirdly, researches on robust fault diagnosis problems based onH∞, filter are performed in time lag systems with nonlinear perturbations. The observer-based robust fault detection filter design problem is formulation as an optimization problem by establishing fault observer and selecting reference model. Linear Matrix Inequality techniques are then applied to prove and gel the existence conditions and the solutions. This designed fault detection filter enable the observation process are not only robust, but also higher sensitivity to faults, so the detection effect is improved. A numerical example illustrates the proposed approach.Finally, the dynamic model is educed. The two aforementioned fault diagnosis methods are put into HVDC system, some familiar faults in transmission of HVDC system are detected. Experiment results show the two methods are applicable to HVDC system.