Hvdc Transmission System Fault Diagnosis Method

As the high-voltage DC transmission (HVDC) system have many advantages which the alternating current transmission system doesn't have, so people pay much attention to the HVDC. The HVDC has risen to the national stratagem. Between "the Eleventh Five-Year Plan" period, China Southern Power Grid focused on building the HVDC project from YunNan to GuangDong, which will be the first±800KV HVDC project in the world. The HVDC project become more and more important in the power system, so the diagnose to HVDC correctly has very important meaning.At present, the fault diagnosis is confronted with some problems, such as non-linear and the absence of failure samples. However, the support vector machine (SVM) has many advantages in solving the problems. Many people focus on it. Wavelet packet transform is a time-frequency localization analysis method which can extract the time-frequency of the signal characteristics effectively. Think of the advantages and disadvantages, we proposes a new method which using the wavelet packet transform to extract the on the failure characteristics of HVDC system. the final support vector machine classification on the feature vector correctly. At last, the fault feature vectors are classified correctly base on SVM.This paper analyzes the reasons why the fault happens in HVDC system and the influence to the whole system, then we constructed simulation models which can reflect many states of HVDC system. We use wavelet packet transform to analyse the DC voltage signal or DC current signal which can reflect the characteristics of fault. We extract effective eigenvectors as the training samples and test samples of SVM. Finally, using the SVM to classification the fault samples, and discuss the SVM parameters which have significant influence on the result of classification.