Sparse random forest-based method for assessing running state of distribution network device

Abstract

The invention discloses a sparse random forest-based method for assessing a running state of a distribution network device. The method comprises the steps of performing correlation analysis on running parameter data and overhaul condition data of the distribution network device in historical data information to obtain an index correlation network of the distribution network device; storing and recording the running parameter data and the overhaul condition data of the distribution network device according to an index correlation characteristic of the distribution network device to form a running state information cluster of the distribution network device; processing the running state information cluster of the distribution network device by utilizing a random forest algorithm, and representing a mapping relationship between the running parameter data and the overhaul condition data of the distribution network device; sparsifying a random forest assessment model of the distribution network device to obtain a sparse random forest-based model for assessing the running state of the distribution network device; and calculating and outputting an assessment result of a to-be-assessed sample by adopting the sparse random forest-based model for assessing the running state of the distribution network device. The method is suitable for efficiently processing high-volume data information and can be used for accurately and efficiently assessing the running state of the distribution network device in service.

Description

technical field [0001] The invention belongs to the technical field of distribution network equipment, in particular to a method for evaluating the operating state of distribution network equipment based on a sparse random forest. Background technique [0002] With the expansion of the scale of the urban distribution network, users have increasingly strict requirements on the reliability of power supply, and the maintenance of distribution network equipment, especially the evaluation of the operation status of distribution network equipment, is becoming more and more important. Traditional distribution network equipment operation status evaluation methods are mainly based on distribution network equipment detection methods such as infrared temperature measurement and patrol inspection, and carry out empowered expert scoring or pattern recognition based on sample learning through the operation status information of its own equipment and fault category information in the mainte...

AI Technical Summary

Problems solved by technology

These two types of traditional distribution network equipment operation status evaluation methods mainly face two technical difficulties: First, due to the variety of distribution network equipment, a large number, and a wide distribution range, the equipment operation status information is huge but redundant. Extremely strong, without considering the factory state parameters of distribution network equipment and the correlation analysis of equipment using correlation analysis, it is difficult for the traditional distribution network equipment operation status evaluation technology to make scientific and reasonable use of existing data

Second, due to the constraints of data information, the traditional sample learning-based distribution network equipment operation status evaluation method tends to rely too much on the relevant data information for a specific type of equipment, resulting in the occurrence of over-fitting phenomenon, and it is difficult to obtain accurate and effective Equipment operating status evaluation results

[0003] To sum up, the contradiction between the existing traditional distribution network equipment operation status evaluation method and the power supply reliability of a large number of distribution network equipment and the shortage of operation and maintenance staff is becoming more and more prominent

Images