CNN-GRU water quality prediction method fusing attention mechanism

Abstract

The invention relates to a CNNGRU water quality prediction method fusing an attention mechanism. The method comprises the following steps of cleaning and normalizing pollutant concentration, river water flow rate and water flow rate data, constructing the data into a vector form, inputting the vector form into a CNN model for feature extraction, and taking feature vectors extracted by the CNN model as input of a GRU network for model training; meanwhile, an Attention mechanism being fused in the model training process to optimize model parameters, and finally, the trained model being used forshort-term prediction of river water quality. According to the method, the strong feature extraction capability of the CNN model, the simple and easy convergence of the GRU network structure and the outstanding important features of the Attention mechanism are fully fused, and accuracy of the model is improved; effectiveness of the method is verified through specific experiments.

Description

technical field [0001] The invention belongs to the technical field of neural network and data mining, and specifically relates to a CNN-GRU water quality prediction method integrated with an Attention mechanism. Background technique [0002] Rivers are an important part of life on earth, providing water for human life, production and agriculture, and are the basis for human survival and development. In recent years, the malicious discharge of a large amount of wastewater has caused more and more rivers to be seriously polluted, which has seriously threatened the ecological balance and sustainable economic development. River water quality prediction is an important means to protect river water environment and deal with river pollution. By predicting the concentration of pollutants in river water resources, it is possible to detect and deal with possible serious pollution events in time, and provide strong data support for the protection of river water resources. In recent ...

AI Technical Summary

Problems solved by technology

[0004] Aiming at the deficiencies in existing river water quality prediction methods, the technical problem mainly solved by the present invention is to use pollutant concentration, river water flow and water flow velocity as model input, use CNN model to fully extract data features, and use GRU network to reduce The complexity of the model, and the integration of the Attention mechanism to optimize the model, improve the prediction accuracy and the robustness of the model

Images