Cleaning method of wind turbine abnormal data

Abstract

The invention belongs to the technical field of data analysis and processing, and particularly relates to a wind turbine generator abnormal data cleaning method based on combination of an optimal intra-group variance and a two-dimensional probability density. The method comprises the steps of: preprocessing wind turbine generator operation data; adopting an optimal intra-group variance method to remove data of the power limiting area; eliminating an abnormal value with sparse density by adopting a two-dimensional probability density estimation method; and obtaining normal operation data through upper and lower boundary lines. By adopting the scheme of combining the optimal intra-group variance and the two-dimensional probability density estimation, the problem that the optimal intra-groupvariance can leave discrete data after cleaning accumulated data is solved, the problem that the two-dimensional probability density estimation cannot eliminate high-density electricity limiting datais solved, and the adaptability of the data cleaning operation condition is improved on the whole.

Description

technical field [0001] The invention belongs to the technical field of data analysis and processing, and in particular relates to a method for cleaning abnormal data of wind turbines based on the combination of optimal intra-group variance and two-dimensional probability density. Background technique [0002] The power curve of a wind turbine is the most important characteristic reflecting its performance. The standard power curve is obtained by testing under standard conditions, but the actual operating conditions of wind turbines are very complicated, and most of them deviate from the standard test conditions. In order to obtain an accurate actual operating power curve of wind turbines, these operating data must be cleaned to eliminate unqualified data. Therefore, the cleaning of wind turbine operation data is one of the core technologies of wind power big data analysis. [0003] There are a large number of operating data lower than the design power of the wind turbine i...

AI Technical Summary

Problems solved by technology

Variance threshold and variance change rate are good for cleaning accumulation-type power curtailment data, but some low-density sparse data will be missed; density-based clustering is good for low-density sparse data cleaning, but high-density power curtailment data cannot be excluded

The main defect of the current wind turbine operation data cleaning technology is poor adaptability to operating conditions

Images