A Soft Sensor for Simultaneous Prediction of Struvite Purity and Recovery Rate

Phosphorus is essential for all living organisms and is widely used in various industries. However, phosphorus resources are nonrenewable and may soon be depleted. Struvite, a high-quality slow-release compound fertilizer containing nitrogen, phosphorus, and magnesium, is a form of phosphorus recovery product. It is typically recovered by applying magnesium salt chemicals to nitrogen- and phosphorus-rich wastewater. Struvite purity and recovery rate are critical quality parameters that dominate its market price and feasibility as a fertilizer. Traditional methods for determining the struvite purity and recovery rate involve complicated sample preparation and analysis processes, which are labor-intensive and would be adverse to the automation of continuous analysis during large-scale struvite recovery. This study presents a new method for simultaneously determining struvite purity and phosphorus recovery by analyzing optical microscope images and combining saturation index information using principal component analysis via MATLAB scripts. The prediction results of this model closely match those of traditional chemical and instrumental analysis methods, suggesting that a reliable model for struvite product quality can be developed by using only physical properties, including optical crystal images and saturation index information. This advancement provides a theoretical foundation for real-time monitoring of the struvite purity.