A data cleaning method based on two-dimensional probability density estimation and a quartile method

Abstract

The invention belongs to the technical field of wind turbine generator data measurement and processing, and particularly relates to a data cleaning method based on two-dimensional probability densityestimation and a quartile method, which comprises the following steps: collecting wind speed, power and pitch angle data at the height of a hub of a wind turbine generator, and performing normalization preprocessing to form a sample data set; Calculating a joint probability density function of the power and the pitch angle by adopting a two-dimensional non-parameter kernel density estimation method, and determining a pitch angle boundary position through a grid division method; Dividing power and pitch angle data into normal data and abnormal data based on the pitch angle boundary line, and obtaining normal wind speed and power data by adopting a data time benchmarking method; And for omission of abnormal data after cleaning, adopting a quartering method to perform data recleaning. The method is high in universality, the data category of the transition area can be effectively identified, a large number of accumulation type power limiting data can be scientifically cleaned, and a reliable data basis is provided for efficiency evaluation, performance analysis, state diagnosis, health management, power prediction and the like of the wind turbine generator.

Description

technical field [0001] The invention belongs to the technical field of wind turbine data measurement and processing, and in particular relates to a data cleaning method based on two-dimensional probability density estimation and quartile method. Background technique [0002] The measured power curve of wind turbines can represent the actual operating conditions of wind turbines, and it is the premise and key to evaluate and predict the power characteristics of wind turbines. The measured power curve of wind turbines is dynamic, affected by a series of factors such as meteorological factors, environmental conditions, system control, equipment failure, etc., the collected wind speed power data has a large number of abnormal data, which cannot be directly used for power curve fitting. Abnormal data cleaning needs to be performed first. Considering the complex and changeable characteristics of the actual power curve, traditional data cleaning methods generally lack versatility,...

AI Technical Summary

Problems solved by technology

Considering the complex and changeable characteristics of the actual power curve, traditional data cleaning methods generally lack versatility, especially for a large number of accumulated power-cut data samples. Misdeletion and missing deletion of abnormal data affect the integrity and correctness of data samples

Images