Work

Genetic Algorithm-Optimized CNN-BiLSTM Framework for Predicting the Remaining Useful Life of IGBT Modules

Year: 2026

Type: article

Abstract: To address the aging and failure issues that arise during the long-term operation of insulated gate bipolar transistors (IGBTs), this paper proposes a method for predicting their remaining useful life (RUL). The proposed method utilizes a genetic algorithm to optimize a hybrid model that combines a convolutional neural network (CNN) with a bidirectional long short-term memory (BiLSTM) network. First, based on the failure mechanism of IGBTs, various commonly used RUL prediction methods are analyzed and compared. Considering that CNNs are particularly effective at extracting spatial features, while LSTMs excel at capturing long-term dependencies in time-series data, a hybrid CNN-BiLSTM model is developed for RUL prediction, with hyperparameters, including the initial learning rate, optimized using a genetic algorithm. Experimental results demonstrate that the proposed CNN-BiLSTM model achieves superior performance across all metrics compared with benchmark algorithms, and the genetic algorithm significantly accelerates the parameter optimization process and enhances the overall training efficiency.

Source: Sensors

DOI: 10.3390/s26061964

Authors: Yukai Hao, Jiao Wu, Zhiheng Zhang, Yuanhao Wang, Tao Wang, Yujie Liang

Institutions: Aviation Industry Corporation of China (China), Xidian University, Guangzhou Institute of Technology

Language: English

Topic: Machine Fault Diagnosis Techniques

Subfield: Control and Systems Engineering

Field: Engineering

Domain: Physical Sciences

SDGs: Industry, innovation and infrastructure, Affordable and clean energy

Open Access status: gold

Estimated APC paid: $3,137