Forecasting of municipal solid waste generation using non-linear autoregressive (NAR) neural models

Area of Science:

• Environmental Science
• Data Science
• Urban Planning

Background:

• Quantifying municipal solid waste (MSW) generation is complex due to multiple influencing variables.
• Accurate MSW estimations are crucial for designing effective waste disposal systems and optimizing waste management strategies.
• Monthly MSW forecasts support budgetary planning and resource allocation for future waste handling.

Purpose of the Study:

• To forecast monthly municipal solid waste (MSW) generation in Nagpur, India, for the year 2023.
• To evaluate the efficacy of non-linear autoregressive (NAR) neural models in predicting short-term MSW fluctuations.
• To identify seasonal patterns and estimate peak and minimum waste generation periods within the forecast year.

Main Methods:

• Development of a classical multiplicative decomposition model with simple linear regression to address data availability issues (max absolute error 6.34%).
• Application of non-linear autoregressive (NAR) neural network models (Model A and Model B) for short-term monthly MSW prediction.
• Utilizing yearly lagged values in NAR model construction to best capture MSW generation variations, achieving high coefficients of efficiency (E=0.99 testing, E=0.97 validation).

Main Results:

• NAR models accurately predicted monthly MSW generation with low absolute maximum errors (Model A: 6.45%, Model B: 3.05%).
• The year 2023 forecast indicates minimum MSW generation in February (39682 ± 471 tons) and maximum generation in December (48504 ± 1569 tons).
• A projected increase of approximately 5345 tons in minimum waste generation is estimated between 2017 and 2023.

Conclusions:

• Non-linear autoregressive (NAR) models are effective tools for short-term monthly municipal solid waste (MSW) generation forecasting.
• Seasonal variations significantly impact MSW generation, with distinct peaks and troughs observed throughout the year.
• Accurate MSW forecasting, as demonstrated in this study, is vital for proactive and efficient urban waste management planning.