Comparison of Particle Filter using SIR Algorithm with Self-adaptive Filter using ARMA for PHM of Electronics

In this paper, an anomaly method has been developed for the prognostication and health monitoring of Electronic assemblies under shock and vibration. Previously, damage initiation, damage progression in electronic assemblies have been monitored using state-space vector from resistance spectroscopy and then be analyzed with particle filter (PF) and the theory of Bayesian. Precise resistance measurement based on the resistance spectroscopy method and the predicted model for the damage process, have been used to quantify the damage initiation and damage progression. However, they vary a lot in different materials and situation. The presented effectiveness of the proposed prognostic health management method based self-adaptive filter and Auto Regressive model. During a shock or vibration test, we can see that the damage of the solder must come from the previous damage in the last state. Therefore, the Auto Regressive model can help us get a precise step propagation function, build the relationship among the continuous state vectors, rate of change of the state vector and acceleration of state vector. With this relationship, we can construct a feature vector. In order to fit different material and situation, the weight of different state variables will be predicted by the self-adaptive filter in which the minimum mean square error algorithm will be used. With the estimated auto-correlation function, cross-correlation function metrics and state parameters, we can propagate the feature state vector into the future and predict the time at which the feature vector will cross the failure threshold. Therefore, remaining useful life has been calculated based on the propagation of the state vector. Standard prognostic health management metrics were used to quantify the performance of the algorithm against the actual remaining useful life.