The subpulmonary ventricular exclusion (Fontan) could effectively improve the living quality for the children patients with a functional single ventricle in clinical. However, postoperative Fontan circulation failure can easily occur, causing obvious limitations while clinically implementing Fontan. The cavopulmonary assist devices (CPAD) is currently an effective means to solve such limitations. Therefore, in this paper the in-silico and in-vitro experiment coupled model of Fontan circulation failure for the children patients with a single ventricle and CPAD is established to evaluate the effects of CPAD on the Fontan circulation failure. Then a sensorless feedback control algorithm is proposed to provide sufficient cardiac output and prevent vena caval suction due to CPAD constant pump speed. Based on the CPAD pump speed-an intrinsic parameter, the sensorless feedback control algorithm could accurately estimate the cavopulmonary pressure head (CPPH) using extended Kalman filter, eliminating the disadvantage for pressure sensors that cannot be used in long term. And a gain-scheduled, proportional integral (PI) controller is used to make the actual CPPH approach to the reference value. Results show that the CPAD could effectively increase physiological perfusion for the children patients and reduce the workload of a single ventricle, and the sensorless feedback control algorithm can effectively guarantee cardiac output and prevent suction. This study can provide theoretical basis and technical support for the design and optimization of CPAD, and has potential clinical application value.