Predicting Escherichia coli levels in manure using machine learning in weeping wall and mechanical liquid solid separation systems

Abstract

An increased understanding of the interaction between manure management and public and environmental health has led to the development of Alternative Dairy Effluent Management Strategies (ADEMS). The efficiency of such ADEMS can be increased using mechanical solid-liquid-separator (SLS) or gravitational Weeping-Wall (WW) solid separation systems. In this research, using pilot study data from 96 samples, the chemical, physical, biological, seasonal, and structural parameters between SLS and WW of ADEM systems were compared. Parameters including sodium, potassium, total salts, volatile solids, pH, and E. coli levels were significantly different between the SLS and WW of ADEMS. The separated solid fraction of the dairy effluents had the lowest E. coli levels, which could have beneficial downstream implications in terms of microbial pollution control. To predict effluent quality and microbial pollution risk, we used Escherichia coli as the indicator organism, and a versatile machine learning, ensemble, stacked, super-learner model called E-C-MAN (Escherichia coli–Manure) was developed. Using pilot data, the E-C-MAN model was trained, and the trained model was validated with the test dataset. These results demonstrate that the heuristic E-C-MAN ensemble model can provide a pilot framework toward predicting Escherichia coli levels in manure treated by SLS or WW systems.