Tidal velocity prediction method based on fractal theory and improved least square support vector machine

Abstract

Aiming at the shortage of the current tidal current velocity research, the invention provides a tidal velocity prediction method based on a fractal theory and an improved least square support vector machine, which has the characteristics that the intermittency and uncontrollability of tidal energy are the main characteristic causing randomness of generator output power, the tidal velocity time series is a nonlinear system with fractal characteristics, by taking historical data as a research object, based on the fractal theory, the Hurst exponent and V statistic of time series are calculated byadopting an R / S analysis method, and the stability and autocorrelation of tidal current are evaluated. A tidal current prediction model is established based on the parameters of the improved least square support vector machine optimized by a dragonfly algorithm for predicting the tidal current. The prediction method can effectively judge the autocorrelation of the tidal velocity, provides the theoretical basis for the prediction of the tidal velocity, and effectively improves the prediction accuracy and the operation efficiency through the improved prediction model.

Description

technical field [0001] The invention relates to a method for predicting short-term tidal flow velocity of an ocean platform based on a least squares support vector machine improved by fractal theory and dragonfly algorithm, and belongs to the field of tidal energy flow velocity prediction. Background technique [0002] Tidal energy is a kind of renewable clean energy, which has been developed on a large scale in my country in recent years. However, renewable energy generally has the characteristics of intermittent, uncontrollable and topological structure diversification, and tidal energy also exists. . Due to the ebb and flow of the tidal flow, the tidal flow velocity varies from time to time, which will cause randomness in the power output of the generator. Accurate tidal energy power prediction can provide an important guarantee for power dispatching and power system reliability assessment, and effectively reduce the impact of tidal energy on the power grid. Since tidal ...

AI Technical Summary

Problems solved by technology

In the process of LSSVM predictive modeling, the parameters that affect the accuracy of the predictive model are mainly the regularization parameter and the width of the kernel function, but these two parameters are generally selected by trial and error, which causes troublesome prediction and large errors, so some scholars propose to use Particle swarm algorithm, genetic algorithm, ant colony algorithm, etc. optimize the least squares support vector machine, but the above-mentioned intelligent algorithm will have slow modeling speed, easy to fall into local optimum, low prediction efficiency and other problems

Images