An online detection method for municipal sludge moisture content based on ultrasonic transmission technology

Accurate, real-time moisture content (MC) detection for municipal sludge is critical for optimizing dewatering and reducing treatment costs, which is difficult to implement due to its extremely complex physical and chemical properties. This study develops a novel online detection method, integrating ultrasonic transmission with a multivariate mixed regression (MMR) model, for non-destructive, high-precision online MC detection. Device geometry was optimized via COMSOL simulation, selecting a 40 kHz emission frequency and an 8 cm container distance, which together balance cavitation effects, energy dissipation, and cost-effectiveness. An experimental device incorporating adaptive density correction and temperature compensation was built, achieving stable measurements with a rapid response (<15 s). The dedicated MMR model was specifically designed for this system's characteristics and rigorously evaluated against multivariate linear regression (MLR) and backpropagation neural network (BPNN) models using identical data. Results demonstrate the MMR model's superiority: achieving an R² of 0.978, MAE of 1.901, and RMSE of 2.233. Compared to the MLR and BPNN models, the MMR model increases R² by 12.08 % and 10.37 %, respectively, while reducing MAE by 54.71 % and 52.91 %, and RMSE by 58.44 % and 56.16 %. The model was further validated using 30 sludge samples from different treatment plants, confirming its robustness and generalizability. This research provides a rapid, accurate, and stable solution for online MC detection, holding significant potential for real-time dewatering process optimization.