Machine learning-assisted assessment of municipal solid waste thermal treatment efficacy via rapid image recognition and visual analysis

Decentralized thermal treatment is a common method for municipal solid waste (MSW) disposal in rural areas. However, evaluating the effect of incineration has always been challenging owing to the difficult and time-consuming measurements involved. Herein, this study presented a rapid image recognition method for assessing the effects of thermal treatment on MSW using a neural network algorithm and a BAEVA 1.0 software based on the relation between the ignition loss of the incinerated bottom ash and its color properties. Through Pearson correlation analysis, the results demonstrated a strong correlation (R² > 0.80) between the ignition loss and the R, G, and B color values. To enhance evaluation accuracy, we introduced the backpropagation artificial neural network (BPANN) algorithm, which exhibited an average evaluation error of only 3.21 in crossvalidation, 27.9 % lower than that of the linear regression model. Building upon the BPANN, we developed BAEVA 1.0 as a software tool for thermal treatment effect evaluation. This tool exhibited advantages in functionality, convenience, and accuracy compared to existing methods. Overall, this research provides an important rapid assessment approach for evaluating the effects of MSW incineration when measurement conditions are unavailable.