Hybrid Imperialist Competitive Algorithm-Artificial Neural Network for waste generation prediction

Nowadays producing municipal solid waste (MSW) is one of the critical consequences of modern life in the world. It is too delicate to evaluate the amount of municipal solid waste generation (MSWG) in future for programming, and managing the systems to reduce, recycle, reuse, or repel solid waste (SW). Developing artificial neural networks (ANN) with evolutionary algorithms is a proper method to achieve precise results in MSWG approximation. Radial basis function (RBF) and Multilayer Perceptron (MLP) are two types of ANN which are used in time series prediction. RBF is a function that its value is based on the distance from a center point and can be used for approximation issues. MLP is a type of ANN which has layers and neurons and, it utilizes supervised learning techniques to solve approximation problems. The imperialist competitive algorithm (ICA) is a new socio-politic algorithm which optimizes RBF and MLP in this study. Structures and practical parameters of each method are introduced, and the results are compared with each other. Results depict ICA has a powerful ability in ANNs optimization and time series prediction reaches high precision, especially in the RBF model. RBF-ICA model achieves a higher correlation coefficient than MLP-ICA one, so it is a stronger model for MSWG forecasting. Correlation Coefficient indicates the similarity of model outputs and real outputs and more significant correlation coefficient means the higher precision of model evaluation ability. This precise method is so useful and practical in urban planning and sustainable development systems.

کلیدواژه ها:

Waste management ، Artificial neural network ، Imperialist competitive algorithm