Xiaoyun Sun

The study of fault diagnosis model of DGA for oil-immersed transformer based on fuzzy means Kernel clustering and SVM multi-class object simplified structure

A model based on fuzzy kernel C-means clustering (FKCM) and support vector machine (SVM) multi-class object simplified structure is proposed for oil-immersed transformer fault diagnosis. The basic idea is, firstly, the training samples are clustered by fuzzy kernel C-means clustering algorithm so as to cancel the isolated data that have no compactness characteristics, then the right ones clustered by fuzzy kernel C-means clustering are put into the classifier of SVM multi-class object simplified structure and trained by this structure. Finally, the fault of the transformer can be detected through the SVM structure. The result shows that the precision is better than the traditional one, and the reliability and effectiveness using above method is satisfied in fault diagnosis.

Application of Combinatorial Probabilistic Neural Network in Fault Diagnosis of Power Transformer

Probabilistic Neural Network (PNN) overcame the shortcomings of entrapment in local optimum, slow convergence rate which was in BP algorithm. With enough training samples, PNN obtained the optimal result of Bayesian rules. Because of the fast training rate, the training samples can be added into PNN at any time. So, PNN is fit to diagnose the fault of power transformer and has auto-adaptability. In order to improve the classification accuracy, the conception of combination is introduced into PNN. The fault diagnosis of power transformer is consisted of 4 Probability neural networks in this paper. PNN1 is used to classify the normal and fault. PNN2 is used to classify the heat fault and partial discharge (PD) fault. PNN3 is used to classify the general overheating fault and severe overheating fault. PNN4 is used to classify the partial discharge fault, and energy sparking or arcing fault. The example shows that the effect of combinatorial PNN is a good classifier in the fault diagnosis of power transformer. The combinatorial PNN has better diagnosis accuracy than BPNN and FUZZY algorithm.

Application of information processing technology on non-destructive detecting of rock bolts

The parameters such as the length of rock bolts, anchorage length, free length or construction defects (e.g., the position of less density in anchorage body) are important to evaluate the quality of rock bolts. Because of the harsh environment, the measured signal includes many kinds of noise, which have a great effect to the signal. It is difficult to detect the signal effectively by traditional information processing technology, and therefore the new information processing technology based on non-destructive detecting is provided. The steps are as follows: first, excite rock bolts using pseudo-random signal, then acquire the reflection signal and transmit to a computer, and remove the reflection signal noise using the Hilbert-Huang transform method and extract the characteristic value, finally, calculate the parameters of rock bolts using cross-correlation method and complete detection of rock bolts. Experimental results show that the effect of the above detection method is better than traditional corre...

An improved method on reducing measurement noise based on Hilbert-Huang transform

For nonstationary signal in rock bolts detecting, an effective denoise method based on Hilbert-Huang transform (HHT) is presented, cross-correlation for denoise signal is used to estimate the length of rock bolts. At first, by empirical mode decomposition (EMD) algorithm, a sum of intrinsic mode functions (IMF) are got, and then, the IMF coefficients in useful signal dominative layer are proceeded using soft threshold method, those IMF coefficients are reconstructed, at last, cross-correlation is used to estimate the length. Compared with the traditional denoise method, the above method is better to remove measurement noise and estimate the length.

Application of Kernel C-Means Clustering and Dempster-Shafer Theory of evidence in power transformers fault diagnosis

Transformers fault diagnosis plays a vital role in running security and reliability. The detected information is collected from the disperse sensor, which is lack of the data fusion analysis and easily lead to decision error and leak. A model of oil-immersed transformer fault diagnosis based on the collaborative method of Kernel C-Means Clustering (KCM) and multi-source information data fusion is presented. The basic idea is that the trained samples are clustered first by using KCM, and then the Dempster-Shafer theory of evidence Fusion method is used to train the chosed sample, and decide the transformer fault. The result of the method shows that the above method can reduce the uncertainty efficiently and have a good performance of transformer fault diagnosis.

A new effective method for removing measurement noise in eddy current nondestructive inspection

For tube detection, an effective denoise method based on Hilbert-Huang Transform (HHT) is presented. At first, by using an empirical mode decomposition algorithm, a sum of Intrinsic Mode Functions (IMF) are obtained; then, the IMF coefficients in useful signal dominative layer are proceede d using soft threshold method; at last, those IMF coefficient s are reconstructed. Compared with the traditional wavelet Transform, the HHT method is better to remove measurement noise.

Application of Learning Vector Quantization network in fault diagnosis of power transformer

Learning Vector Quantization (LVQ) network is presented to analysis the fault of power transformer. The oil gases extracted from transformer oil form the input vector of LVQ network. The connection weights vector is determined with teacher guide. Compared with radius function neural network (RBFNN), LVQ network is easy to perform with high efficiency. In order to improve the classification accuracy, the conception of combination is introduced. The fault diagnosis of power transformer is consisted of 4 LVQ networks. The first LVQ network is used to classify the normal and fault. The second LVQ network is used to classify the heat fault and partial discharge (PD) fault. The third LVQ network is used to classify MC-overheating faults in magnetic circuit and EC-overheating faults in electrical circuit. The fourth LVQ network is used to classify RSI-discharge faults related to solid insulation, USI-discharge faults unrelated to solid insulation. By comparing with the RBF neural network algorithm for the same 122 input set, we conclude that the LVQ network a good classifier for the fault diagnosis of power transformer.

Application of Data Mining Technology Based on FRS and SVM for Fault Identification of Power Transformer

Data mining (DM) technology based on Fuzzy Rough Set (FRS) and Support Vector Machine (SVM) are presented to classify the Fault of power transformer. Improper or inadequate Dissolved Gases Analysis (DGA) data may lead to failure fault classification of power transformer. SVM, through statistical learning theory, provides a way of classification information by generating optimal kernel based representative DGA data. In order to make full use of the classification ability of SVM and improve the fault classification accuracy, FRS is used to pre-classify the transformer fault and the multi-level power transformer fault diagnosis model based on FRS and SVM was presented in this paper. By comparing with the traditional method like neural network, there is less fault data discriminated by FRS and SVM model and the accuracy for power transformer fault diagnosis is improved.

The study of transformer fault diagnosis based on means kernel clustering and SVM multi-class object simplified structure

A model based on means kernel clustering and support vector machine (SVM) multi-class object simplified structure is proposed for transformer fault diagnosis. The basic idea is, firstly, the training samples are clustered by means kernel clustering algorithm, then the right ones clustered by means kernel clustering are put into the classifier of SVM multi-class object simplified structure and trained by this structure, finally, the fault of the transformer can be detected. The result shows that the precision is better than the traditional one, and the reliability and effectiveness using above method is satisfied in fault diagnosis.

The study of fault diagnosis model of DGA for oil-immersed transformer based on SVM active learning and K-L feature extracting

A model based on support vector machine (SVM) active learning and Karhunen-Loeve(K-L)feature extracting is proposed for oil-immersed transformer fault diagnosis, and a SVM active learning algorithm with the Euclidian distance based on Mercer function is introduced to select the training sample data. The K-L transform is used to extract the characteristics of the sample data set, and the sample data set that has reduced six dimensions to three dimensions is showed in the three-dimensional figure. The SVM active learning algorithm is used to select and classify the fault types. The result shows that the precision is better than the traditional one, and the reliability and effectiveness using above method is satisfied in fault diagnosis.