WOA-QRLSTM reservoir inflow probabilistic prediction method based on mean shift clustering

Abstract

The invention discloses a WOA-QRLSTM reservoir inflow probabilistic prediction method based on mean shift clustering. The WOA-QRLSTM reservoir inflow probabilistic prediction method comprises the steps: 1, preprocessing collected reservoir inflow and influence factors thereof; 2, clustering the preprocessed data set by using a mean shift clustering algorithm, and dividing the data into a training set and a test set; 3, putting training set data into the WOA-QRLSTM prediction model of the quantile regression long and short term neural network optimized by the whale algorithm for training, putting test set data into the trained WOA-QRLSTM prediction model, and obtaining reservoir inflow prediction values under different quantiles; and 4, calculating the probability density of the reservoir inflow flow in the future according to the reservoir inflow flow predicted values under different quantiles through kernel density estimation. According to the invention, the accuracy of reservoir inflow prediction can be improved, so that effective reservoir inflow prediction information is provided for reservoir operation scheduling.

Description

Technical field [0001] The invention belongs to the field of reservoir storage flow prediction, specifically a WOA-QRLSTM reservoir-based flow rate probability prediction method based on mean drift clustering. Background technique [0002] Reservoir storage flow prediction affects the operation scheduling arrangements and water resources of the reservoir, is an important decision-making basis for the operation of the reservoir; and the upstream water, rainfall, climate change will affect the reservoir storage Flow prediction results, how to improve the precision and stability of forecasts are important directions of research. Compared to the traditional reservoir storage flow point prediction method, the probability reservoir's storage flow prediction method can better reflect the uncertainty of the warehousing flow of the reservoir, thus providing more scientific and reasonable scientific scheduling for the reservoir storage flow. in accordance with. [0003] Traditional histori...

AI Technical Summary

Problems solved by technology

[0003] Traditional forecasting models based on historical data, such as multiple linear regression models, artificial neural network models, support vector machine algorithms, etc., use a single forecasting model to describe the relationship between reservoir inflow flow and the relationship between reservoir inflow flow and factors. Overall, the forecasting effect is not ideal, and the forecasting accuracy is not high, so it is difficult to obtain more accurate prediction results of reservoir inflow flow

Images