Method and device for anomaly monitoring of power plant production subsystem based on two-stage clustering

Abstract

The invention relates to an abnormality detection method and a device of a power plant production subsystem based on two-stage clustering, belonging to the technical field of thermal power generation.The method comprises the following steps: extracting monitoring data of a plurality of measurement points related to the power plant production subsystem to be measured, and obtaining a time data sequence of each measurement point; The outlier time of each measuring point is obtained by the first stage clustering evaluation. By carrying out the second stage clustering evaluation, the data of themeasuring points and the time period corresponding to the core cluster points of the clustering are obtained, and the abnormal situation of the production subsystem of the power plant is determined. The invention starts from the unsupervised learning, does not need a large amount of professional knowledge and experience, and only learns from the recent historical data and cooperates with a small amount of business experience, so that the fault position and corresponding time can be quickly determined. Moreover, the abnormal false positive rate is reduced by using multiple clustering.

Description

technical field [0001] The invention relates to the technical field of thermal power generation, in particular to a two-stage clustering-based abnormality detection method and device for a production subsystem of a power plant. Background technique [0002] At present, the abnormal detection of domestic power plants mainly comes from the following two aspects: [0003] 1. Fault analysis for specific components, such as vibration monitoring for steam turbines and fans; leak detection for boiler tubes; analysis for boiler life. Anomaly detection and fault diagnosis are only carried out for local equipment, lack of overall grasp of the operating system, and the method has poor scalability. [0004] 2. For the analysis of a large number of specific fault samples, it is necessary to sort out a large amount of fault experience data and establish a knowledge base or case set before abnormal detection based on rules or machine learning can be carried out. Fault diagnosis system an...

AI Technical Summary

Problems solved by technology

In daily equipment maintenance work, there are many types of power plant equipment failures, and the failure repetition frequency is low

The phenomenon of equipment failure will also change with the change of working conditions, it is difficult to establish a complete expert experience database

Images