A machine learning fan status monitoring method and device based on cloud computing platform

Abstract

The invention discloses a cloud computing platform-based wind generator state monitoring method and apparatus adopting machine learning. The method comprises the steps of setting a hierarchical extreme learning machine model structure to obtain a first hidden layer initial output matrix H; performing compression processing on the first hidden layer initial output matrix H by using a compressed sensing method; uploading compressed data to a cloud computing platform through the internet of things; restoring the compressed data uploaded to the cloud computing platform; finishing state monitoringmodel training of a hierarchical extreme learning machine method; and according to wind generator data collected in real time, finishing monitoring of real-time states of all subsystems of a wind generator. By utilizing a data compression characteristic of compressed sensing, the quantity of the data uploaded to the cloud computing platform is reduced; and the first hidden layer initial output matrix of a hierarchical extreme learning machine is uploaded only, so that prediction model structure and parameters are ensured to be secure. Compared with a conventional extreme learning machine method, the hierarchical extreme learning machine method has higher monitoring precision.

Description

technical field [0001] The invention relates to a remote state monitoring system, in particular to a machine learning wind turbine state monitoring method and device based on a cloud computing platform. Background technique [0002] This year, in order to reduce greenhouse gas emissions and environmental pollution, my country has built a large number of large-scale wind power plants. However, areas rich in wind resources are often located in mountains, deserts, or seas; these areas often have harsh weather conditions and are not suitable for operators to be on duty for a long time, so it is very necessary to install a remote status monitoring system for wind turbines to monitor the operating status of wind turbines in real time . A typical large-scale megawatt wind turbine often needs to monitor more than one hundred state variables at the same time. According to different variable types, the signal sampling frequency is between 100KHz-1Hz. This means that a large amount o...

AI Technical Summary

Problems solved by technology

[0004] However, using the cloud computing platform requires uploading monitoring data to the cloud through the Internet, and data calculation and storage are implemented in the cloud, which brings the risk of data information leakage, and the cost of using the cloud computing platform depends on the amount of uploaded data

Therefore, traditional machine learning methods cannot meet the needs of practical engineering applications

Images