Abstract:The insulated gate bipolar transistor (IGBT) is widely used in the power electronic system, but its aging state is difficult to predict in advance due to the complicated failure mechanism, which will affect the performance of the equipment, and even cause serious disaster. Therefore, we propose a combined prediction model based on adaptive VMD decomposition (AVMD) and attention-based gated recurrent unit (GRU-AT) to achieve accurate prediction of the aging state of IGBT. For the nonlinear characteristics of the IGBT failure characteristic parameters, AVMD with the key parameters optimized by improved sparrow search algorithm (ISSA) was used to disaggregate the character sequence to a series of finite wide subcomponents, which overcome the interference of the irregularity of aging time sequence data to the prediction accuracy. Secondly, the attention mechanism is introduced to capture the temporal feature relationship between the current moment output and the historical degraded data, further improving the generalization ability of the model. Finally, the GRU-AT network is used to model each wide modal sub-component independently, and final prediction results of IGBT aging parameters are obtained by superimposing the prediction results of each mode. The experimental results indicate that the AVMD-GRU-AT model designed in this article has superior prediction performance in both over-the-top single-step prediction and multi-step prediction of aging state of IGBT.

What problem does this paper attempt to address?

### The Problem Addressed by the Paper This paper aims to address the challenge of predicting the aging state of Insulated Gate Bipolar Transistors (IGBTs). Due to the widespread application of IGBTs in power electronic systems, their aging state is difficult to predict accurately in advance, which not only affects equipment performance but can also lead to severe catastrophic failures. Therefore, the authors propose a combined prediction model based on Adaptive Variational Mode Decomposition (A VMD) and Attention Mechanism Gated Recurrent Unit (GRU-AT) to achieve precise prediction of IGBT aging states. ### Specific Problems 1. **Complex Failure Mechanisms**: The aging process of IGBTs has complex failure mechanisms, making it difficult to accurately predict their aging state. 2. **Nonlinearity and Non-stationarity**: The aging characteristic parameters of IGBTs exhibit significant nonlinearity and non-stationarity, posing challenges to the accuracy of prediction models. 3. **Limitations of Existing Methods**: - **Model-based Methods**: Such as analytical models and physical models, although they can fit the relationship between equipment life and experimental data, they ignore the failure mechanisms and processes of the modules. - **Data-driven Methods**: While they can avoid complex parameter calculations and variable acquisition, a single model is difficult to learn the characteristics of aging when dealing with complex non-smooth signals, and they lack generalization ability, resulting in low prediction accuracy. ### Solution The A VMD-GRU-AT model proposed by the authors addresses the above problems through the following steps: 1. **Adaptive Variational Mode Decomposition (A VMD)**: - Using an improved Sparrow Search Algorithm (ISSA) to optimize the key parameters of VMD, the aging characteristic sequence is decomposed into multiple finite bandwidth sub-components, overcoming the impact of irregular aging time series data on prediction accuracy. 2. **Attention Mechanism**: - Introducing an attention mechanism to capture the temporal feature relationship between the current output and historical degradation data, further improving the model's generalization ability. 3. **Gated Recurrent Unit (GRU)**: - Using the GRU-AT network to independently model each wide mode sub-component, and obtaining the final IGBT aging parameter prediction results by superimposing the prediction results of each mode. ### Experimental Results Experimental results show that the A VMD-GRU-AT model designed in this paper exhibits superior prediction performance in both single-step and multi-step predictions of IGBT aging states.

Research on IGBT aging prediction method based on adaptive VMD decomposition and GRU-AT model

An Analytical Approach for IGBT Life Prediction Using Successive Variational Mode Decomposition and Bidirectional Long Short-Term Memory Networks

IGBT Fault Prediction Combining Terminal Characteristics and Artificial Intelligence Neural Network

Prediction of IGBT Gate Oxide Layer's Performance Degradation Based on MultiScaleFormer Network

Research on IGBT Lifetime Prediction Based on Residual BP Neural Network

IGBT Aging Monitoring and Remaining Lifetime Prediction Based on Long Short-Term Memory (LSTM) Networks

Self-attention-based adaptive remaining useful life prediction for IGBT with Monte Carlo dropout

Performance Degradation Modeling and Its Prediction Algorithm of an IGBT Gate Oxide Layer Based on a CNN-LSTM Network

Degradation state estimation for insulated gate bipolar transistor based on multi-scale fusion learning

Predicting aging of IGBT solder layer using saturation voltage approach with CPO-SVR data modeling

RUL Predict of IGBT Based on DeepAR Using Transient Switch Features

Remaining Useful Life Estimation of Insulated Gate Biploar Transistors (igbts) Based on a Novel Volterra K-Nearest Neighbor Optimally Pruned Extreme Learning Machine (VKOPP) Model Using Degradation Data.

Genetic Neural Network Prediction on Medium and High Voltage IGBT Switching Performance

Fault prediagnosis of power electronic devices in urban new energy system

An Aging-Degree Evaluation Method for IGBT Bond Wire with Online Multivariate Monitoring

Reliability evaluation of IGBT power module on electric vehicle using big data

Novel Critical Gate-Based Circuit Path-Level NBTI-Aware Aging Circuit Degradation Prediction.

A Lifetime Prediction Method for IGBT Modules Considering the Self-Accelerating Effect of Bond Wire Damage

Research on Fault Prediction of Power Devices in Rod Control Power Cabinets Based on BiTCN-Attention Transfer Learning Model

A screening method of retired insulated gate bipolar transistor for reuse based on multiple indicators and integrated model

Research on Power Device Fault Prediction of Rod Control Power Cabinet Based on Improved Dung Beetle Optimization–Temporal Convolutional Network Transfer Learning Model