The high-frequency electromechanical impedance (EMI) method creates active motivation by piezoelectric transducers (PZT) affixed to the surface of the structure, and monitor the structure's health by continuous monitoring and analyzing changes of admittance signals of PZT. However, the EMI method is susceptible to environmental and operational variations, leading to false alarms of structural damage. In light of this problem, the impact of the structural workload on the electromechanical impedance signatures is eliminated by the time series co-integration method. This method is based on the non-stationary characteristics of the PZT impedance spectrum admittance signal under the dynamic load of the structure. The non-stationary time series of the impedance spectrum under the influence of the dynamic load is transformed into a stationary time series by linear combination and the health of the structure is determined according to the co-integration margin sequence. The experiments of loose bolt damage identification of aluminum beam structure under dynamic stress are carried out to verify the effectiveness of the proposed method. The results show that the method can isolate damage-sensitive features from stress variations to detect the existence of bolt loosening. The EMI co-integration method can eliminate the impact of load on structural health monitoring and time to identify structural damage.