Water quality time sequence prediction method based on improved Seq2Seq framework

Abstract

The embodiment of the invention discloses a water quality time sequence prediction method based on an improved Seq2Seq framework, the improved Seq2Seq framework is adopted, a constructed model has good prediction performance on time sequence data, a known time step and a prediction time step can be freely set, and the prediction length is flexible and variable by using coding and decoding processes. By adopting the method, the key problem of predicting a single variable by using a covariable in the water quality prediction problem is solved, the GRU model is used as an encoder and a decoder, and the FM model is integrated, so that the problems of high sparseness and high-dimensional feature interaction of data are solved, and the technical problem that a water quality prediction model in the prior art is not solved is solved.

Description

technical field [0001] The invention relates to a water quality detection method, in particular to a water quality time series prediction method based on an improved Seq2Seq framework. Background technique [0002] In recent years, with the continuous rise of the industrial level, the number of factories has also continued to increase. The improper treatment of waste materials and wastewater by some factories is the biggest cause of water pollution in various regions. Once the regional water quality is polluted, the surrounding animals and plants will be damaged. The living environment will also be seriously threatened, and will eventually endanger the ecology of the entire water area. Therefore, the governance of the water environment has become more and more important, and in the governance of the water quality environment, the prediction of water quality data plays a pivotal role. If the pollution can be dealt with in a targeted manner before it occurs, it will make the g...

AI Technical Summary

Problems solved by technology

(2) The sampling frequency of water quality data is not fixed

The defect of the above two schemes is that CNN is very sensitive to sparse data, and the data sets in the real world are often not perfect, and various null values ​​or singular values ​​are prone to appear, which will seriously affect the ability of CNN to obtain feature information

Images