Utilization of process network synthesis and machine learning as decision-making tools for municipal solid waste management

Abstract

This paper introduces the utilization of two different optimization models as a decision-making tool for municipal solid waste management. This study aims to provide a framework of decision-making in managing municipal solid waste in Malaysia and demonstrating the function model's potential for a decision-making tool. Process network synthesis is used as a first optimization model to generate feasible pathways for municipal solid waste conversion technologies. A standard data mining technique using machine learning was applied for a simple case study of municipal solid waste management. Data were divided with a ratio of 90:10 for training and testing data and further analysis using different model functions. The model functions involved are linear regression, multilayer perceptron neural network, and sequential minimal optimization regression. The linear regression function provides a better correlation coefficient than other models, 0.5894. The correlation coefficient for testing data slightly increases. The multilayer perceptron model improved its correlation coefficient when inputting the testing dataset. The multilayer perceptron model's correlation coefficient reached 0.7169 compared to the linear regression function. Therefore, this case study used the multilayer perceptron model as a basis. Results showed that both integrations of optimization models could be successfully used as a basis in waste management decision-making tools with good performance. The decision-making tool provides a faster and efficient way for the user and stakeholder to manage municipal solid waste. Developing a decision-making tool framework for municipal solid waste management can help achieve better municipal solid waste management.