A New Approach for the Optimization of Biowaste Composting Using Artificial Neural Networks and Particle Swarm Optimization

A novel approach to optimize the composting process of biowaste (BW) mixed with sugarcane filter cake (SFC) and the product quality was attempted in the present study by adopting Artificial Neuronal Network (ANN) and the particle swarm optimization (PSO) algorithm. The effectiveness of the co-composting process depends on operational parameters such as Mixing Ratio (MR) and Turning Frequency (TF). Using the optimization of these factors, the process time can be reduced while product quality can be maximized. This study includes the simultaneous evaluation of both operational parameters, with SFC being the amendment material (BW:SFC MR of 90:10, 80:20 and 70:30) and a TF of 1, 2 and 3 turnings per week. The simultaneous effect of the two operational parameters was evaluated using a central composite design. ANN was used to predict the behaviour of the response parameters, and the PSO algorithm was used to optimize the process and the final product quality. The results of the simulations with ANN suggest that the BW:SFC ratios of 81:19 and 75:25, with a TF of two times/week and an estimated operation time of 76–94 days, correspond to a final product with the most adequate physicochemical quality for agricultural use. The optimization with PSO showed the optimal local at a BW:SFC MR of 76.9:23.1 with a turning frequency of two times weekly. An 80-day process is recommended to optimize the final product quality. The model can be useful to define design criteria and operational conditions during biowaste composting.