For finding the optimized allocation of sensor to monitor status and fault diagnosis for large process industry system, an optimization method of sensor allocation based on multilevel flow model (MFM) is proposed. An updated method of FMEA is presented based on the systematic MFM. An optimization model of sensor allocation is setup with a basic constraint function by considering the detection capacity and the differential ability of the fault and a new composite optimized target function of parameters is the sensibility of testing point fault, reliability & cost factor of the fault. In addition, the corresponding algorithm is created. Finally, the case of power plant deaerator shows that the optimization model of sensor allocation is effective. It also provides a new way to study the optimized allocation of sensors for large process industry system with the target of fault diagnosis.