Research on Circuit Performance Degradation Prediction Method Based on SSA-LSTM & GRU

During the actual operational phase, electronic systems typically go through a series of intermediate degradation states from their initial normal state until a fault occurs. By utilizing degradation data to predict changes in the operational state, it becomes possible to take proactive measures before the electronic system's performance reaches the failure threshold, thereby preventing unexpected situations. A novel performance degradation prediction method called SSA-LSTM & GRU is introduced to overcome the shortcomings of accuracy and efficiency in single model performance degradation prediction. The method starts by applying singular spectrum analysis (SSA) reconstruction and decomposition to the data to extract the relevant components. Subsequently, an adaptive weighting module is incorporated, combining the long and short term memory (LSTM) network with the gated recurrent unit (GRU) network, resulting in the SSA-LSTM & GRU performance degradation prediction method. Experimental results validate the effectiveness and feasibility of this approach, demonstrating its superior prediction accuracy compared to single model methods.