Hidden Markov Model for Municipal Waste Generation Forecasting

Accurate municipal waste generation forecasting can provide theoretical guidance for disposal capacity design in waste management systems to achieve sustainable urban development. Although a few approaches have been proposed for solving the problem, they are still hard to measure and trace fluctuations dynamically. In this paper, we propose a model-driven statistical learning method, which is Hidden Markov Model based on Gaussian Mixture Model, to trace waste generation dynamically for both small and large dataset scenarios. Then, both the Expectation Maximization algorithm and the Viterbi algorithm are employed for parameters learning and finding out the most probable sequence of hidden states respectively. Finally, two case studies test a small dataset using Shanghai solid waste generation, and a large dataset using wastewater generation of a Spanish urban sewage treatment plant. The computational results demonstrate that the proposed approaches are effective to solve the municipal waste generation forecasting problem.