Hydropower station equipment state-based data prediction model subdivision method

Abstract

The invention relates to the field of hydropower station data management, and aims to provide a hydropower station equipment state-based data prediction model subdivision method, which is characterized in that machine learning is applied to construct a model conforming to a hydropower generating set equipment operation rule, and the model comprises a known fault class, a normal operation class, a regular event class and an accidental event class; and according to product continuity data characteristic requirements, data is subjected to time sequence enriching processing, and effective support is provided for enriching modeling factors.

Description

technical field [0001] The invention relates to the technical field of hydropower station management, in particular to a data prediction model subdivision method based on equipment status of a hydropower station. Background technique [0002] A hydropower station is a comprehensive engineering facility that converts water energy into electrical energy. Generally, it includes reservoirs and water diversion systems of hydropower stations, power plants, and electromechanical equipment formed by water-retaining and water-discharging structures. The high water level of the reservoir flows into the powerhouse through the water diversion system to drive the hydroelectric generator set to generate electricity, and then enters the power grid through the step-up transformer, switch station and transmission line. [0003] In terms of existing products, we have investigated Huake Tongan TN8000 hydropower unit condition monitoring system, Aojiyi HPU2100 vibration swing monitoring system...

AI Technical Summary

Problems solved by technology

At present, although domestic equipment status monitoring and analysis systems have been widely used in domestic hydropower enterprises, these systems basically do not fully use the complete monitoring data of the equipment itself to conduct self-learning by computers. The self-modeling machine learning algorithm only supports Real-time status monitoring, unable to achieve early warning

Since the operating conditions of hydropower station equipment are very complex and the boundary conditions are volatile, it is impossible to judge new abnormalities or failures, which often result in missed reports. These abnormalities or failures are precisely what we are most concerned about and the most fatal; due to some abnormalities or failures Although the event is obtained by some sensors, it is caused by abnormal equipment in other parts, so false alarms will be generated

Images