Inverter Commutation Failure Study In HVDC/AC Hybrid System

With the development of nationwide power grid interconnection, more and more HVDC projects have been putting into commercial operation recent years. Long distant bulk power transmission induced security problem of AC/DC interconnected system is of the most concern. Commutation failure (CF) is a dynamic perturbation in HVDC system. It is most often caused by ac fault induced voltage reduction. Normal operation of the valves will be achieved again in some cases, but continuous or repetitive CF may cause HVDC system block or even further deterioration of the whole system. Voltage collapse and control-induced voltage oscillations existing in combined form in ac/dc system can lead to persistent CFs. It's very important to study CF in HVDC system and take measures in advance to avoid catastrophic consequence.This paper discusses the microcosmic process of Commutation. Reverse recovery phenomenon of thyristor has been studied. Current variation during and after commutation is given in sectional functions.The main reason of CF is commutating voltage decrease caused by disturbances in ac system connected to inverter. Ac system strength can be indicated by short circuit ratio (SCR). Moreover, CF is related to dc current, firing angle, extinction angle and the reactance of converter transformer. There are also other factors such as zero-crossing point shift of commutating voltage and ac frequency spectrum. TianGuang HVDC project is used to illustrate the affect of these factors on CF onset.This paper selects the suitable wavelet for CF signal wavelet transform. The different properties of wavelets are analyzed in detail. It is shown that the orthogonality, support, regularity, moments, and symmetry of a wavelet function may affect the CF detection to some extent, and the critical factor is the moments of the mother wavelet. Discrete Wavelet Transform has executed respectively on normal sinusoidal voltage signal and alternative voltage record with CF in Tianguang HVDC project. The transform result shows that db10 is the cogent wavelet.This paper proposes a novel approach of CF recognition using wavelet transform and Shannon entropy technique. Only voltage signal at inverter is used as the input. Ac voltage is decomposed by DWT to get details and approximations on all scales. Local modulus maxima of high frequency detail coefficients are calculated to locate the fault time. Feature vector is extracted from the details and approximation of fault signal, and standardization is carried out by comparison to that of the corresponding normal signal. CF can be identified by comparing the Euclidean distances between feature vectors. The proposed approach is verified by fault signals simulated in a complete 12-pulse HVDC transmission system in MATLAB/SIMULINK. The results have approved its feasibility and preciseness.HVDC control strategy and its application in GaoZhao HVDC transmission system in China Southern Power Grid have been studied. HVDC control can prevent subsequent CF and help system recovering after faults, but can not significantly affect the onset of initial CF. Three phase to ground fault in ac system electrically near the inverter may cause ac voltage drops. Careful comparison is carried out between Constant Current Control mode (CCC) and Constant Power Control mode (CPC). System stability simulations are performed using NETOMAC program. The results show a highly uniform while dc system in CCC or CPC mode with optimal controller parameters. The digital simulation also shows that HVDC modulation can augment system stability.