Order Publications

Reference: ISH2017_399

Type:
ISH Collection
Title:

Evaluation of power transformer through information fusion of multi-dimensional data

 

Abstracts

Accurate evaluation of the states of power transformer is based on the analysis of vast scale multi-dimensional examining data. Such data include traditional test data, which are in form of digit numbers and are relatively easy to process. Other data are hard to crunch, such as the test reports scanned into computer system as a picture, the infrared thermographs presenting temperature information. During the evaluation process, different test items give totally different ‘states’ of the same transformer. How to find the comprehensive results from the conflicting data remains a big challenge. This article has addressed above problem with big data analysis technology based on real online monitoring data and off line test data of over 50 transformers. A model that consists of the fuzzy set theory, fuzzy analytical hierarchical process and modified D-S evidence theory is employed to assess the condition of the transformers. Firstly, the fuzzy set theory is used to analysis the principal factors of the detection data, the detection data has four factors: dissolved gas analysis, electrical testing, oil testing, and miscellaneous factors. Next, the inputted condition parameters indices are processed by the membership functions of a fuzzy distribution to obtain relative impairment grades which represent the relative grade of the transformer condition towards the fault. Then the fuzzy analytical hierarchical process is introduced to calculate the weights of indices and factors, the proposed method can better solve the uncertainty existed in the comparison matrix by the experts. Finally, the modified D-S evidence theory is employed to revised the weights and achieve the goal of multi-dimensional data fusion and to avoid the conflicts of different data. The proposed method has been applied successfully. The results indicate that the evaluation method can reflect the current state of a transformer quantitative. Meanwhile the method can provide a better maintenance strategy and early warning of possible failures.
 

File Size: 1,1 MB

Pages NB: 6

Year: 2017

 
 
 
Download
Member
FREE free
Non member
FREE free
 

Your AccountYour Account

Password forgotten

Members

may download free of charge all publications including most recent ones


Non Members

may download free of charge publications over three years old and purchase all publications

For any questions in connection with the on-line library, contact:


Publications and Editions Manager

publications@cigre.org