Investigating the Ability of Support Vector Machine and Wavelet Transform Method in Predicting Water Quantity and Quality (Case Study: Anzali Lagoon)

Articles in Press

Document Type : Original Article

Authors

1 PhD student, Department of Water Science Engineering, Lahijan Branch, Islamic Azad University, Lahijan, Iran.

2 Assistant Professor, Department of Agriculture, Lahijan Branch, Islamic Azad University, Lahijan, Iran

10.22034/iwrr.2024.431495.2734

Abstract

In this research, in order to prepare a numerical prediction infrastructure of the state of quantity and quality changes of Anzali wetland, prediction using artificial intelligence method. In this research, due to the necessity of accurate measurement of climate forecasts from existing quantitative and qualitative statistics of surface currents, from field data and ground data in a 20-year period from 1999 to 2020, based on calculations with Processing was used in the software environment. The results indicated that the regression extracted with the RBF function had a high match compared to the linear regression. Also, after confirming the experimental method using the SVM model, a wavelet transform model was developed to determine the final parameter of CWT. The results showed that the value of the target function in the data range of -2 to 2 in the radial function and the linear model was almost close to each other in the range of -0. 9 to 0. 1, but these numbers were different in the polynomial function, which results it indicated a high compatibility of the regression extracted with the RBF function compared to the linear regression. Also, the results showed that the SVM prediction model well fitted the RBF function on the data in accordance with the linear regression fitting in the experimental method of discovering the trend and the time series of the user's data. The results in this case indicate CWT with a density of four in the recorded periods in accordance with the images.

Keywords

Main Subjects

Iran-Water Resources Research

Articles in Press, Accepted Manuscript
Available Online from 10 March 2024

Files

History

  • Receive Date: 20 December 2023
  • Revise Date: 04 March 2024
  • Accept Date: 10 March 2024

Share

How to cite

Statistics