River water quality prediction method considering space-time correlation and meteorological factors

Abstract

The invention relates to a river water quality prediction model considering space-time correlation and meteorological factors. Firstly, a spatial attention module is embedded into an encoder to extract significant spatial correlation between upstream and downstream water qualities and interaction between water quality indexes; secondly, the meteorological factors and the spatial features are inputinto a decoder, and a time attention module is embedded into the decoder to extract important time sequence features; and finally, multi-step prediction of the water quality indexes is achieved by using a long-short-term memory neural network. The model has the advantages that the implicit space-time dependence relationship between upstream and downstream water qualities and the interaction between water quality indexes can be mined, and the defect that a traditional water quality prediction model is difficult to explain can be overcome.

Description

technical field [0001] The invention relates to the technical field of water quality forecasting, in particular to a river water quality forecasting method taking into account time-space correlation and meteorological factors. Background technique [0002] In recent years, water eco-environmental problems have seriously threatened the sustainable development of my country's economy and the safety of domestic water for urban residents. Building a high-precision water quality prediction model is conducive to understanding the internal change rules and temporal and spatial evolution trends of river water quality. Prevention provides the basis for decision-making. [0003] At present, many scholars have studied the methods of water quality index prediction, but there are still the following deficiencies: 1. These existing prediction methods are limited to single-site prediction, ignoring the influence of upstream water quality on the downstream water quality change trend; 2. Diff...

AI Technical Summary

Problems solved by technology

[0003] At present, many scholars have studied the methods of water quality index prediction, but there are still the following deficiencies: 1. These existing prediction methods are limited to single-site prediction, ignoring the influence of upstream water quality on the downstream water quality change trend; 2. Different locations The impact of upstream sites on downstream water quality will also be different. There are two situations of basic stability and sudden change in water quality at different times. These two change modes also have different impacts on future water quality conditions.

Images