Dip et al. (2026) A Hybrid PSO-SVM Framework Incorporating Advanced Preprocessing Techniques for Enhanced Prediction of River Water Level

Identification

• Journal: Lecture notes in civil engineering
• Year: 2026
• Date: 2026-01-01
• Authors: M. Mubtasim Fuad Dip, Sajal Kumar Adhikary, Shuvendu Pal Shuvo, Md. Jobayer Parvez Ratul, Usmi Akter
• DOI: 10.1007/978-3-032-18708-6_30

Research Groups

Department of Civil Engineering, Khulna University of Engineering & Technology, Khulna, Bangladesh

Short Summary

This study develops a hybrid Particle Swarm Optimization-Support Vector Machine (PSO-SVM) framework, integrated with Hodrick-Prescott (HP) filtering and Empirical Mode Decomposition (EMD) preprocessing techniques, to enhance the prediction accuracy of short-term river water levels, especially in noisy data environments. The framework demonstrates high potential for improved predictions, achieving R² values of 0.9 and 0.89 for HPF-SVM and EMD-SVM, respectively.

Objective

• To develop a hybrid PSO-SVM framework incorporating advanced preprocessing techniques (Hodrick-Prescott filtering and Empirical Mode Decomposition) to minimize noise in river water level data and enhance prediction accuracy, addressing limitations of classical machine learning approaches.

Study Configuration

• Spatial Scale: Pussur River, Hiron Point station, Bangladesh.
• Temporal Scale: 1 year of hourly (1 h) water level data.

Methodology and Data

• Models used: Hybrid Particle Swarm Optimization-Support Vector Machine (PSO-SVM), Hodrick-Prescott Filter (HPF), Empirical Mode Decomposition (EMD). The final models are HPF-SVM and EMD-SVM.
• Data sources: Observed river water level data from the Pussur River.

Main Results

• The HPF-SVM model achieved a coefficient of determination (R²) of 0.9 for river water level prediction.
• The EMD-SVM model achieved a coefficient of determination (R²) of 0.89 for river water level prediction.
• Both hybrid models demonstrated good prediction capabilities for short-term river water levels, particularly effective with noisy data.

Contributions

• Development of a novel hybrid PSO-SVM framework that integrates advanced signal processing techniques (HPF and EMD) to effectively handle noisy river water level data.
• Demonstrates significant improvement in short-term river water level prediction accuracy compared to classical machine learning approaches, as evidenced by high R² values.
• Provides a robust methodology for enhanced water level forecasting in environments characterized by data noise.

Funding

[No specific funding information was provided in the paper text.]

Citation

@article{Dip2026Hybrid,
  author = {Dip, M. Mubtasim Fuad and Adhikary, Sajal Kumar and Shuvo, Shuvendu Pal and Ratul, Md. Jobayer Parvez and Akter, Usmi},
  title = {A Hybrid PSO-SVM Framework Incorporating Advanced Preprocessing Techniques for Enhanced Prediction of River Water Level},
  journal = {Lecture notes in civil engineering},
  year = {2026},
  doi = {10.1007/978-3-032-18708-6_30},
  url = {https://doi.org/10.1007/978-3-032-18708-6_30}
}