Prediction of moisture content in municipal sludge drying process based on GA-BP-Garson model

During the municipal sludge drying process, the detection of internal moisture poses challenges. To accurately predict the variation pattern of internal moisture during the hot air drying process of municipal sludge, Back Propagation (BP) neural network and GA-BP (Genetic Algorithm Back Propagation) neural network algorithms were utilized to establish moisture prediction models for the municipal sludge hot air drying process, with drying time, drying temperature, sludge layer thickness, and flow differential pressure as input variables, and moisture content as the output variable. Moreover, a sensitivity analysis was conducted on the GA-BP neural network to investigate the impact of the four input variables on the prediction outcomes. The results indicated that the coefficients of determination (R²) for the BP and GA-BP moisture prediction models on the test set were 0.999 55 and 0.999 64, with root mean square errors (RMSE) of 0.513 17 and 0.455 23, respectively, demonstrating that the GA-BP algorithm-based prediction model achieved better performance, accurately predicting the dynamic changes of moisture during the municipal sludge drying process. The sensitivity analysis revealed that drying time had the most significant impact on the GA-BP moisture prediction model. These findings could provide a theoretical basis for optimising sludge drying processes and procedures and offer a reference for the resource utilization of sludge.