Executive Summary : | Model Predictive Control (MPC) is the most widely used multi-input, multi-output, advanced control strategy in industrial applications. Being a model based technique, the performance of MPC depends on the accuracy of the model. For most practical applications, due to uncertain model form, uncertain parameter values, measurement noise, and unmeasured disturbances, system-model mismatch is inevitable. Stochastic Model Predictive Control (SMPC) considers model uncertainties in the optimized control input calculation, thereby improving the control performance of MPC for systems under operational uncertainties. However, SMPC does not guarantee recursive feasibility or is intractable. Scenario-based SMPC (SB-SMPC) helps overcome the feasibility problem of SMPC. However, SB-SMPC is computationally expensive due to the need of generating scenarios at each time instant and including them in the optimal control problem (OCP). Multiparametric programming can help improve the computational efficiency of SB-SMPC. Multiparametric MPC (mpMPC) is an explicit control strategy that helps shift the OCP offline thereby reducing online solution to point location and function evaluation. The decision variables are the control inputs. The system states and outputs are the uncertain parameters. The novelty of the work lies in developing multiparametric scenario-based stochastic model predictive control (mpSB-SMPC) algorithm by also considering the scenarios as uncertain parameters and including them in the OCP. The reformulation of SB-SMPC to solve the OCP explicitly using multiparametric programing will be the key contribution of this work. The developed mpSB-SMPC algorithm will be tested on a building energy management system under operational uncertainties. A first principles dynamic model of a test house to predict indoor temperature and pollution concentrations as a function of indoor conditions, outdoor conditions, and control actuator settings will be developed. Closed-loop simulations will be performed under the developed mpSB-SMPC, conventional MPC, and conventional SB-SMPC controllers. The advantages and limitations of the developed mpSB-SMPC method will be studied. The computational and control performance will be compared against conventional controllers. |