Numerical robustness and efficiency of generalized predictive control algorithms with guaranteed stability

Three previous papers proposed modifications to the generalized predictive control algorithm which guarantee closed-loop stability. The first two adopted the philosophy of constrained receding horizon predictive control (CRHPC), whereas the third adopted a stable generalized predictive control (SGPC) strategy: first stabilize then control the plant. Here the authors examine the relationship between CRHPC and SGPC and show that theoretically the two approaches are equivalent, but that CRPHPC could be subject to significant numerical instability problems. Two improved implementations of CRHPC are proposed, but SGPC is shown to have better numerical stability and efficiency.<>