Index self-optimization prediction method and device based on antecedent relationship

Abstract

The invention discloses an index self-optimization prediction method and device based on an antecedent relationship. The method comprises the following steps: formulating a collaborative adjustment rule, calculating time-delay collaborative correlation to form a time-delay correlation analysis table, and screening time sequence indexes to form an antecedent analysis table; if the time duration of the indexes is far smaller than the number of the indexes, re-sampling the index set, and carrying out stepwise regression; constructing an index importance measurement formula, and performing self-optimization on the preceding index set to form a key preceding index set; and stepwise regression is carried out by using the key preceding index set, and parameter estimation and prediction are carried out. According to the index self-optimization prediction algorithm based on the antecedent relationship, the problem of prediction failure caused by the fact that the index dimension is far higher than the index number in the high-dimensional data analysis process is solved.

Description

technical field [0001] The invention relates to the technical field of regular volatility data processing, in particular to an index self-optimization prediction method and device based on antecedent relationship. Background technique [0002] Large-scale wind farms usually occupy a large area, so the inspection and maintenance of wind turbines requires a lot of manpower and material resources. In order to improve the daily maintenance efficiency of wind power plants, it is necessary to construct a statistical prediction method based on wind turbine operation feedback indicators, conduct a comprehensive analysis, and identify high-risk wind turbines for key maintenance. That is, in order to optimize the daily maintenance of wind farms, it is necessary to construct a high-dimensional predictive analysis method based on the operation status of each component of the wind turbine system to predict the possibility of wind turbine failure. [0003] For newly installed wind turbin...

AI Technical Summary

Problems solved by technology

[0003] For newly installed wind turbines, it has the characteristics of short observation period and many operation feedback data indicators, and the existing statistical prediction methods may not be able to give accurate and robust prediction results

To this end, we provide a self-optimized index prediction method based on the prior relationship to solve the problem of prediction failure caused by the index dimension being much higher than the number of indexes in the process of high-dimensional data analysis

Images