This paper presents an integrated model based upon the fuzzy approach and evidential reasoning decision-making approach to condition assessment of power transformers. An assessing index system, including the DGA data, electrical testing, and oil testing, is established to facilitate the assessing model. The model is composed of two levels, that is, the fuzzy model and evidential reasoning model. The fuzzy model is proposed for generating the original basic probability assignments for the second-level model. Afterwards, an evidential reasoning decision-making model is utilized to combine all of the evidence and give an overall evaluation conclusion. Based upon this integrated model, a decision-making procedure is put forward to serve as an effective tool for transformer condition assessments. The results show that the assessing model is capable of offering an overall evaluation of the observed transformer condition.

In this paper, a set of dissolved gas analysis (DGA) new feature combinations is selected as input from the mixed DGA feature quantity, and an improved krill herd (IKH) algorithm optimized support vector machine (SVM) transformer fault diagnosis model is established to solve the problem that the single characteristic gas or characteristic gas ratio which are utilized as the DGA feature quantity cannot fully reflect the transformer fault classification. The following work has been done in this paper: (1) IEC TC 10 fault data and other 117 sets of fault data in China are preprocessed in order to reduce the influence on diagnosis results causing by the edge data in the fuzzy area. (2) The SVM parameters and 11 features are encoded by a binary code technique. (3) A preferred DGA feature set for fault diagnosis of power transformers is selected by genetic algorithm (GA) and SVM. (4) GA is utilized to optimize the parameters of SVM. Combining with cross-validation principle, a transformer fault diagnosis model based on IKH algorithm to optimize support vector machine is established. The fault diagnosis results based on the new fault sample show that the proposed DGA feature set increase the accuracy by 26.78% and 10.83% over the DGA full data and IEC ratios. Moreover, the accuracy of IKHSVM is better than the GASVM, Back-Propagation neural network (BPNN) and particle swarm optimization optimized support vector machine (PSOSVM), the accuracy rate are 85.71%, 75%, 64.29% and 71.43% respectively which proves the validity of the proposed fault diagnosis model.

Moisture is one of the critical factors to determine the service life of transformers. The moisture inside the transformer oil-immersed insulation could be quantified with feature parameters. This paper proposes and develops a Genetic Algorithm Support Vector Machine (GA-SVM) model to carry out the moisture diagnosis. Present findings reveal that these feature parameters can be obtained by using frequency domain spectroscopy. Therefore, a novel model for predicting frequency domain spectroscopy curves is firstly reported based on a small number of samples, which could be utilized to obtain the feature parameters database to develop GA-SVM. Then, the moisture diagnosis in lab and field conditions is presented to verify its feasibility and accuracy. The novelty of this work is in an exploration of the reported model, as an intelligent based moisture diagnosis tool for power transformers.

The thermographic diagnostic of substation insulators can facility to detect their insulation status in time. However, the detection effect of infrared insulator images is usually poor due to the complex background of substation scene images. In order to solve this problem and improve the automatic detection capability of substation equipment, an infrared insulator image detection model based on the improved feature fusion single shot multibox detector is proposed in this paper. The model combines multi-scale feature maps to generate a new feature pyramid, and designs a new feature enhancement module in the shallow network of the model to improve its ability to extract features of infrared images of substation insulators. Meanwhile, the clustering algorithm is used to calculate the target aspect ratio information of the dataset to realize the adaptive change of the aspect ratio of the default box. Furthermore, the transfer learning is employed to further improve the learning efficiency of the model. Before the formal experiment, the design scheme of feature enhancement module is confirmed by ablation studies. Experiments are performed on the collected infrared insulator images, instrument transformer images and Pascal VOC 2007 dataset. It is demonstrated experimentally that this model achieves excellent detection results.