Water quality prediction method combining divide-and-conquer method and water quality periodicity

Abstract

The invention discloses a water quality prediction method combining a divide-and-conquer method and water quality periodicity, and belongs to the technical field of water quality prediction. The water quality prediction method aims to solve the problem that complex change rules of water quality data cannot be sufficiently learnt by a single predictor and accordingly prediction accuracy is poor, a problem decomposition solving frame based on data driving is adopted in the method, the frame can automatically divide problem space according to data characteristics in the learning process, the number of predictors does not need appointing by staff, and better robustness is achieved. Predictor effect evaluation and water quality data periodicity are combined, so that change characteristics of water quality data on the time scale are well utilized, and a more accurate problem decomposition algorithm is obtained. According to frame leaning content, data to be predicted are mapped to a predictor group correspondingly having prediction advantages, a prediction result is obtained through weighting of corresponding predictors, and better accuracy and stability are achieved.

Description

technical field

[0001] The invention belongs to the technical field of water quality forecasting, and relates to a water quality forecasting method combining divide and conquer method and water quality periodicity. Background technique

[0002] Water quality prediction is an important part of water resource management. Precise prediction of water quality changes can realize early warning of pollution events such as water quality deterioration and algal bloom outbreaks, and provide reference for relevant decision-making. At present, water quality prediction methods are mainly divided into two categories. The first type is based on the water flow, physical, chemical and other factors in the water quality mechanism, such as QUASAR and WASP. Mechanism-based models comprehensively consider various environmental factors affected by water quality, and have been applied in many watersheds, but they require relatively complete observation samples and prior knowledge, and are limite...

Images