A Novel Method for Remaining Useful Life Prediction of RF Circuits Based on the Gated Recurrent Unit-Convolutional...

Area of Science:

• Electrical Engineering
• Artificial Intelligence
• Reliability Engineering

Background:

• Remaining Useful Life (RUL) prediction is crucial for RF circuit reliability.
• Data-driven methods offer an effective alternative to traditional approaches, bypassing complex failure mechanism knowledge.

Purpose of the Study:

• To propose a novel RUL prediction method for RF circuits using a gated recurrent unit-convolutional neural network (GRU-CNN).
• To enhance the accuracy and reduce the uncertainty in RUL predictions for RF circuits.

Main Methods:

• Data normalization for improved algorithm efficiency.
• Hybrid health score calculation using Euclidean and Manhattan distances to evaluate circuit degradation.
• RF circuit life cycle segmentation based on hybrid health scores.
• RUL prediction using the GRU-CNN model at each life cycle stage.

Main Results:

• The proposed GRU-CNN model achieved a Root Mean Square Error (RMSE) that is 3/5 of that of standalone GRU and CNN models during the normal operation stage.
• The GRU-CNN model significantly minimized prediction uncertainty.

Conclusions:

• The GRU-CNN model demonstrates superior performance in RUL prediction for RF circuits compared to traditional GRU and CNN models.
• The proposed method effectively enhances RF circuit reliability through accurate RUL prediction.