Method for evaluating health state of power transformer under incomplete multi-modal information

Abstract

The invention discloses a method for evaluating the health state of a power transformer under incomplete multi-modal information, and the method comprises the steps: obtaining the parameter information of the power transformer, and enabling the parameter information to comprise a plurality of feature indexes; collecting fault cases of the power transformer, and performing discretization processing on each modal feature according to the difference that whether multiple feature indexes of the power transformer with different voltage grades, rated capacities and operating years exceed threshold requirements or not, and establishing an information decision table of each modal feature by using a rough set; adopting a PSO algorithm to solve the NP-Hard problem of attribute reduction under the multi-modal background, and obtaining the kernel feature of the health state of the power transformer; using a TFG-T task flow mechanism to complete kernel feature integration; and evaluating the state of the power transformer through the integrated kernel features and the information decision table. The method is used for solving the problem of state evaluation of equipment under incomplete information, fusing the multi-modality of detection data, and providing technical support for intelligent operation and maintenance and digital transformation of a power transformer.

Description

technical field [0001] The invention relates to the technical field of transformer state evaluation, in particular to a method for evaluating the health state of a power transformer under incomplete multimodal information. Background technique [0002] With the continuous deepening of the digital transformation of the power system, more and more applications of sensor networks and intelligent measurement technologies have been applied, which provides strong data support for the implementation of accurate assessment of equipment status under panoramic monitoring data. As a core device in the power system, the power transformer has a direct impact on whether the power system can run safely and stably. [0003] "Should be repaired and must be repaired" and "intelligent maintenance" have become the inevitable trend of equipment maintenance. When evaluating the status of power transformers, the index system is complex and the data types are diverse, showing multi-modal character...

AI Technical Summary

Problems solved by technology

[0005] 1) The current state evaluation guidelines still rely on traditional artificial knowledge and experience, and most of them adopt the method of threshold value discrimination. The evaluation process is greatly affected by the subjective, and it is difficult to objectively reflect the accurate law between faults and characteristics;

[0006] 2) Relevant research focuses on the complementarity of each characteristic index, and obtains the subjective and objective weights of the index through methods such as correlation and analytic hierarchy process, so as to obtain the comprehensive health status of power transformers through weighted analysis. The integrity of the indicators is high, the information coverage is small, the sample is not representative enough, and the data input required by the model is high, complete and standardized;

[0007] 3) The industry's method for evaluating the state of power transformers has changed from the initial single-parameter classification evaluation to the method of multi-parameter interaction, taking into account the transformer's operating environment, bad working condition information and other factors. Currently, the key evaluation indicators for power equipment There is still less research on construction methods;

[0008] 4) In the traditional power industry mode, most of the characteristic data reflecting the operating health status of power transformers are scattered and stored in various subsystem databases with different logical entities or data characteristics, and the problem of information islands is prominent

Images