A newly early warning model for anaerobic digestion systems: Based on an improved sparrow search algorithm combined with least square support vector machine

Anaerobic digestion as an important means of organic waste treatment will play a key role in the realization of ecological civilization and the goal of double carbon. However, the instability of the system due to the high sensitivity to operating conditions restricts the economy and sustainability of the current commercial biogas projects in China. Machine learning as an early warning and control tool for many industrial systems is also applicable to anaerobic digestion systems. Existing studies focus on the biogas or methane yield prediction of the system, while there are few studies have considered the acid-bases indicators, which is crucial to the stability of the system. In this study, an improved sparrow search algorithm was developed, and after comparing its performance with selected optimization algorithms using CEC2017 test suite, combined with LSSVM, was applied to the prediction of eight different indicators of anaerobic systems, and eight datasets were validated. The results show that the optimization algorithm proposed in this study improves the performance of LSSVM and the model of ISSALSSVM shows excellent potential in the early warning and controlling of anaerobic digestion system.