An Automatic Abstract Classification Method for Defect Data Based on Bayesian Network

Abstract

A method for automatically summarizing defect data based on Bayesian networks, including: a. Integrating the text information in each quadrant of defect appearance, defect location, defect description, defect equipment, and defect cause in each defect data, and integrating The final defect record is taken as a text analysis object; the defect text sample is segmented using the Chinese word segmentation system; b. Using the Bayesian classification algorithm, the defect sample data is analyzed in three dimensions: equipment name, defect location, and defect type. Classify to obtain the classification type of some defect dimensions; c. According to the relationship between the various dimensions within the defect data, use the classified part of the defect dimension abstracts to construct a Bayesian network model of defect abstract information to obtain the learning rules of the defect model ; d. Automatically digest and classify the actual defect data, so as to standardize the defect data and provide basic data for the analysis and application related to equipment defects.

Description

Technical field: [0001] The invention relates to the data processing technology of electric power equipment, in particular to an automatic abstract classification method for defect data based on Bayesian network. Background technique: [0002] The data information contained in the equipment defect data itself is very rich, and the record of the defect data is artificially filled in the form. Different staff members have different ways of thinking about defect judgments, resulting in defect data entered into the system showing colloquial information, lack of information, and vague expressions. Under the background of rapid increase in data scale and intricate data structure, dig out valuable patterns and laws in data resources, guide the operation of power equipment, and assist in risk monitoring of equipment. Therefore, according to the status quo of actual defect data, seeking a more efficient and scientific method for processing equipment defect data, constructing a Bayes...

AI Technical Summary

Problems solved by technology

Different staff members have different ways of thinking about defect judgments, resulting in defect data entered into the system showing characteristics such as colloquial information, lack of information, and vague expressions.

Images