Identification method for photovoltaic power abnormal data

Abstract

The invention provides an identification method for photovoltaic power abnormal data. The identification method comprises the steps of fitting an irradiance-photovoltaic power Copula function parameter by utilizing photovoltaic power station irradiance and photovoltaic power measured data; establishing a probability power curve based on the relationship of the two random variables, namely, the irradiance and the photovoltaic power both described by the Copula function; summarizing characteristics of abnormal data points and establishing an abnormal data criterion through an irradiance-photovoltaic power scatter diagram; identifying and screening abnormal data based on the Copula function and the abnormal data criterion, and establishing a new data set; if the abnormal data are identified, repeating the steps mentioned above after the abnormal data are removed, and carrying on identifying a new data set; and if no abnormal data are identified, identifying the abnormal data in the original data set directly by utilizing the criterion and the criterion power curve. The identification method for photovoltaic power abnormal data is applicable for identifying photovoltaic power abnormal data of various photovoltaic power stations, has universality and can solve the problem of high proportion of the abnormal data in the original data.

Description

technical field [0001] The invention belongs to the field of power system new energy power generation. In particular, it relates to a method for identifying abnormal photovoltaic power data under the condition that the photovoltaic power data contains a high proportion of abnormal data. Background technique [0002] Accurate and reliable photovoltaic power time series data are the basis for photovoltaic power generation performance analysis and power forecasting. However, the quality of power data collected on-site by many photovoltaic power plants is poor, which greatly hinders the information mining and in-depth application of these data. There are many reasons for generating abnormal photovoltaic power data, such as communication failures, equipment abnormalities, artificial power cuts, etc. Among them, the problem of abnormal photovoltaic power data caused by artificial power cuts is particularly serious in my country. A high proportion of abnormal photovoltaic power d...

AI Technical Summary

Problems solved by technology

Existing methods mostly rely on the detection of module status, such as screening abnormal data based on the relationship between module temperature and operating voltage, which requires detection of operating voltage. Pay more attention to abnormal data at the station level in forecasting

[0004] However, the existing methods cannot fully adapt to the actual situation of photovoltaics in my country; assuming a certain probability density function of photovoltaic power distribution, such as using the 3-sigma principle to identify abnormal data, according to such assumptions, the photovoltaic power in each irradiance interval The distribution laws are independent of each other, but in fact, if the irradiance and photovoltaic power are two correlated random variables, the power distribution laws in each irradiance interval are not independent, and if the independent processing does not conform to the actual law , so that the ability to identify abnormal data in practical applications is limited; the formulation of simple anomaly identification rules adopts the method of rule identification, on the one hand, it relies too much on empirical laws, and on the other hand, it does not consider the correlation between photovoltaic power and main influencing factors. The recognition effect is not good in practical applications; it is less targeted at the problem of high proportion of abnormal data. Factors such as power cuts and equipment failures lead to a high proportion of abnormal data in photovoltaic power data. In practical applications, a high proportion of abnormal data will lead to statistical analysis results Deviating from the real situation, making the misidentification rate of abnormal data high

Images