Degradation Prediction of Proportional Solenoid Based on Data–Model Interactive Method

Proportional solenoids are widely used as key power conversion devices in advanced servo systems. The proportional solenoids are highly prone to inter-turn short circuit faults after prolonged operation due to thermal stress and insulation degradation, leading to gradual performance degradation or even catastrophic system failure. However, predicting the degradation of proportional solenoids within embedded packages remains a significant technical challenge due to limited accessibility. In this paper, a novel data-model interactive degradation prediction approach is proposed, which eliminates the requirement for additional sensors or signal injection. A physics-based degradation model is developed to characterize the internal degradation process under thermal stress. To this end, the particle filtering method is first employed to estimate the unmeasurable states of the solenoid using indirect sensor measurements from the solenoid valve. Then, the expectation maximization method is applied to identify the degradation-related hidden parameters in the physics-based degradation model, thereby enabling accurate degradation prediction. Experimental validations are conducted on a proportional solenoid test rig under diverse operating conditions. The experimental results demonstrate that the proposed method significantly improves the degradation prediction accuracy compared with the state-of-the-art algorithms.