Water quality space-time correlation prediction method based on long and short term memory and radial basis function neural network

Abstract

The invention relates to a water quality space-time correlation prediction method based on long and short term memory and a radial basis function neural network. A water quality space-time fusion prediction model based on LSTM-RBF is provided by considering space-time characteristics and multivariate correlation of water quality changes. The method is mainly divided into five processes. The method comprises the following steps: 1) cleaning data, and respectively constructing a time data set and a space data set; 2) performing grey correlation analysis on the time data set obtained in the step 1), and performing spatial feature dimension reduction by using depth self-coding; 3) constructing an LSTM water quality time dimension prediction model according to the parameters after feature extraction in the step 2) to obtain a time dimension prediction result; 4) taking a result after feature dimension reduction in the step 2) as input of a spatial dimension RBF model, establishing an RBF spatial dimension water quality prediction model, and obtaining a spatial dimension prediction result; and 5) taking the time prediction result and the space prediction result as two new features, and fusing by using the model tree to obtain a time-space fusion water quality parameter prediction result.

Description

Technical field: [0001] The water quality prediction method of the present invention is specifically applied to the prediction of important pollution parameters in the water environment. Aiming at the time characteristics of water quality changes, a method based on long short-term memory (Long Short-Term Memory, LSTM) and radial basis function (Radial Basis) is proposed. Function, RBF) neural network water quality spatiotemporal correlation prediction method. Aiming at the multiple correlations of water quality changes, gray relational analysis and deep self-encoding dimensionality reduction are used to select time and space dimension features respectively. This method can effectively improve the accuracy and generalization ability of the water quality prediction model. Background technique: [0002] Water quality prediction is an important support for water pollution control and the basis of water environment management. By analyzing the water quality prediction results, ...

AI Technical Summary

Problems solved by technology

When it comes to practical applications, the accuracy of a single water quality prediction model is not high, and it cannot make full use of the spatiotemporal characteristics and multiple correlations of water quality changes

Images