It is difficult for a single feature to comprehensively reflect the degradation fault of electromagnetic relay. A fault detection method of electromagnetic relay is proposed based on time parameter fusion by selecting the pick-up time, release time and super-path time as multiple fault characteristic variables according to the fault mechanism of relay. Firstly, In consideration of the fuzzy uncertainty of the degraded fault, the fuzzy membership function is constructed according to the fuzzy threshold of the feature, and the membership degree of the data to the working state of the relay is obtained and expressed as fault evidence, and then the evidence reasoning algorithm is applied to the fusion of the multi-feature fault evidence to get the relay fault information. Finally, the fault detection results are obtained by iteratively fusing the historical, current and predicted fault information, and adaptively adjust the reliability of the fault information, taking into account the reliability and sensitivity of the detection results. The experimental results show that the proposed method is 5.2% in mistake report rate, 2.0% in failure to report rate, has significant advantages in accuracy and anti-interference ability, and can effectively improve the performance of electromagnetic relay degradation fault detection.