Vol.:(0123456789)
Clean Technologies and Environmental Policy
https://doi.org/10.1007/s10098-024-03091-8
ORIGINAL PAPER
Forecasting municipal solid waste generation using advanced
transformer and multi‑layer perceptron techniques
Neeraj Kumar
1
· P. Rajeswari
2
· D. Jeya Priya
3
· M. Uma Maguesvari
4
Received: 24 July 2024 / Accepted: 11 November 2024
© The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2024
Abstract
Municipal Solid Waste generation forecasting is crucial for the treatment process, policy decision-making, and waste manage-
ment. The challenges of accurately predicting municipal solid waste generation have become increasingly crucial for effective
urban waste management. Therefore, this paper presents a novel Lightweight GraphFormer model designed to enhance both
the accuracy and efficiency of Municipal Solid Waste generation predictions by integrating temporal and spatial dependen-
cies within waste data. This proposed approach is designed by combining a graph graph-based transformer model with a
Lightweight Multi-Layer Perceptron. Statistics and World Bank Database is used to collect Municipal solid waste data and
is preprocessed to handle the missing data in the database by using the imputation approach. Further, the embedded data are
fed into the long short-term memory model for effectively learning and extracting temporal features. The objective of the
graph-based transformer model is to extract potential spatial dependence and dominant attention. The novel Multi-Layer
Perceptron such as Precurrent Multi-Layer Perceptron is used within the transformer model that perceives temporal locality
features by fusing feature representations from different time steps. The performance evaluation is conducted through dif-
ferent time series data analyses and quantitative analyses that demonstrate better forecasting performance in municipal solid
waste generation. The time series data analyses are conducted for different types of wastes such as textile, garden, metal, food,
glass, plastic, paper, and others. The quantitative analysis provided that the proposed model attained better performances
of 98.76%, 0.982, and 0.078 from forecasting accuracy, correlation coefficient, and RMSE respectively than other existing
research articles. These results illustrate the model’s potential to significantly enhance the field of waste management by
providing improved predictive capabilities, while maintaining low computational complexity.
* Neeraj Kumar
neerajkumar29@acm.org
1
School of Information Technology, University Teaching
Department, Rajiv Gandhi Proudyogiki Vishwavidyalaya,
Bhopal, India
2
Department of Artificial Intelligence and Data
Science, Panimalar Engineering College,
Poonamallee, Chennai 600123, India
3
Department of Information Technology, St. Joseph’s College
of Engineering, OMR, Chennai 600119, India
4
Department of Civil Engineering, Rajalakshmi Engineering
College, Chennai 602105, India