LSSVML Hammerstein Identification and Nonlinear Model-Based Controller Paradigms for a Real-Time pH Neutralization Process

pH neutralization is a highly nonlinear and complex process in chemical reactors. The pH of untreated industrial wastewater affects product quality, soil, and water. Due to its nonlinear characteristics and complexity, pH control is challenging in the process industry. Experimental studies have shown that identifying pH neutralization systems and designing model-based controllers are effective. In this work, we apply an efficient nonlinear prediction model we developed, Least Square Support Vector Machines-Laguerre Hammerstein (LSSVMLH). The VAF (Variance Accounted For) is 92.8186%, and the MSE (Mean Squared Error) is about 0.5406, indicating high confidence in pH neutralization. The identified LSSVMLH model is used to develop Model-Based Controllers (MBCs), namely Hammerstein Nonlinear Internal Model Controller (HNIMC) and Hammerstein Nonlinear Model Predictive Controller (HNMPC) for real-time control of a pH neutralization process. The HNIMC solves the model inversion problem by discarding all positive zeros outside the unit circle to stabilize the system. The experimental results and comparative analysis show that HNMPC and HNIMC have lower ISE and IAE than LMPC and IMC-based Proportional-Integral-Derivative (PID) controllers. The experimental results demonstrate the effectiveness of the model-based controllers in real-time control of a highly nonlinear pH neutralization process.