In this paper, a strategy of an active control accommodation is proposed to deal with the regulation of SISO plants with large parametric uncertainties. The model of the plant is partitioned into a family of a finite number of LTI models. Then, the associated controllers are designed such that they achieve the same performances for a given control objective. The selection of the adequate controller is based on an online detection algorithm of these models. The stability and bumpless transfer issues linked to the switching of controllers are discussed. The results of a real-time experimentation of this strategy on a laboratory thermal process demonstrate its effectiveness.