Latent-variable Nonlinear Model Predictive Control Strategy for a pH Neutralization Process

Linear model predictive control (MPC) is a widely-used control strategy in chemical processes. Its extension to nonlinear MPC (NMPC) has drawn increasing attention since many process systems are inherently nonlinear. When implementing the NMPC based on a nonlinear predictive model, a nonlinear dynamic optimization problem must be calculated. For the sake of solving this optimization problem efficiently, a latent-variable dynamic optimization approach is proposed. Two kinds of constraint formulations, original variable constraint and Hotelling T-2 statistic constraint, are also discussed. The proposed method is illustrated in a pH neutralization process. The results demonstrate that the latent-variable dynamic optimization based the NMPC strategy is efficient and has good control performance.