Prediction of water quality based on artificial neural network with grey theory

Abstract

In this paper, the grey theory, three type of artificial neural network (back-propagation neural network, radial basis function neural network, and generalized regression neural network) and their combination were used to predict the pH values in the evaluation of water quality. Based on the measured data from the Xielugang in Jiaxin with the post-hoc analysis for the c and p values of the prediction, the results showed that the prediction by using the generalized regression neural network has the averaged relative error 0.61%, and c <0.65, p>0.7.

A Review of the Artificial Neural Network Models for Water Quality Prediction

“…Table 4 summarizes the data set sizes of feedforward and recurrent neural networks involved in this review. According to Table 4, the number of samples applied for water quality prediction varies from 28 [39] to 45,594 [78] which illustrates the fact that ANN models are capable to deal with different size of the dataset. However, there has been no research studying the optimal amount of data required for each ANN model.…”

“…In contrast, feed-forward neural network can predict water quality with relatively little data. In addition to being able to perform prediction tasks, GRNN is also suitable for small data sets (28,32,61,151,159, 265 samples) compared with other types of ANNs [24,[39][40][41][42][43], so researchers should pay some attention to it. The artificial neural network has been widely used in water quality prediction.…”

“…To overcome the local minimum problems [5,18,37,38] GRNNs A modified form of the RBFNNs model There is a pattern and a summation layer between the input and output layers Solving the small sample problems [24,[39][40][41][42][43] WNNs Wavelet function replace the linear sigmoid activation functions of MLPs Solving the non-stationary problems [16,44] ELMs The structure of ELMs is similar to the MLPs Only need to learn the output weight…”

“…Results demonstrated the effectiveness of the proposed methods. Researchers tended to divide the training set data into 70% to 90% of the total data [39,42,49,52,72,[120][121][122][123][124][125][126][127]. Iglesias et al [35] divided the data into training (90%) and testing sets (10%).…”

Xây dựng mô hình dự báo BOD5 cho hạ lưu sông Sài Gòn – Đồng Nai dựa trên các mạng nơ–ron nhân tạo MLP và RBF

“…Bước này thường được thực hiện bằng cách thử và sai để đảm bảo rằng mô hình có thể đạt được hiệu quả tối ưu [13]. Sau khi thực hiện, bộ dữ liệu được chia thành hai phần phục vụ cho quá trình huấn luyện và kiểm tra, cụ thể 75% (185 dữ liệu/1 thông số) được sử dụng cho quá trình huấn luyện và 25% (61 dữ liệu/1 thông số) được sử dụng cho quá trình kiểm tra mô hình, tỷ lệ này cũng đã được áp dụng và đạt hiệu quả cao trong một số nghiên cứu ứng dụng mô hình trí tuệ nhân tạo để dự báo chất lượng nước như nghiên cứu [14][15][16].…”