Scholarlink

<p>Maintaining the reliability of power converters in wind energy systems is crucial for ensuring uninterrupted energy generation, making precise fault diagnosis essential to prevent system failures. Traditional fault diagnosis methods often struggle with handling complex fault scenarios, especially those involving multiple or mixed faults, presenting a significant challenge in the field. Recent advancements in artificial intelligence have provided new solutions to these challenges. This paper introduces a deep learning model that combines a capsule neural network (CapsNet) for capturing and extracting information data with bidirectional long short-term memory (BiLSTM) for enhanced fault identification. The proposed CapsNet-BiLSTM model is applied to diagnose faults in power converters within PMSG-based wind energy system under different scenarios, including simple faults, multiple faults on both the generator and grid sides, and mixed faults between these sides. A comparative analysis is conducted using several recent existing models to assess and justify the effectiveness and performance of the proposed method. Evaluation metrics include precision, recall, f1-score, accuracy, and training efficiency. The results demonstrate that CapsNet-BiLSTM outperforms other models, achieving a testing accuracy of 99.98% for fault diagnosis of power converters, highlighting its effectiveness in fault detection and diagnosis applications within wind energy systems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Fault diagnosis of power converters in PMSG-based wind energy conversion system using capsule network and bidirectional long short-term memory

  • 期刊论文

  • 发表日期：

    2025年5月14日

查看全文
  • Ahmed Mesai Belgacem,
  • Mounir Hadef,
  • Abdesslem Djerdir,
  • Youcef Bekakra,
  • Enas Ali,
  • Salah K. Elsayed,
  • Sherif S. M. Ghoneim,
  • Saad A. Mohamed Abdelwahab

摘要

Maintaining the reliability of power converters in wind energy systems is crucial for ensuring uninterrupted energy generation, making precise fault diagnosis essential to prevent system failures. Traditional fault diagnosis methods often struggle with handling complex fault scenarios, especially those involving multiple or mixed faults, presenting a significant challenge in the field. Recent advancements in artificial intelligence have provided new solutions to these challenges. This paper introduces a deep learning model that combines a capsule neural network (CapsNet) for capturing and extracting information data with bidirectional long short-term memory (BiLSTM) for enhanced fault identification. The proposed CapsNet-BiLSTM model is applied to diagnose faults in power converters within PMSG-based wind energy system under different scenarios, including simple faults, multiple faults on both the generator and grid sides, and mixed faults between these sides. A comparative analysis is conducted using several recent existing models to assess and justify the effectiveness and performance of the proposed method. Evaluation metrics include precision, recall, f1-score, accuracy, and training efficiency. The results demonstrate that CapsNet-BiLSTM outperforms other models, achieving a testing accuracy of 99.98% for fault diagnosis of power converters, highlighting its effectiveness in fault detection and diagnosis applications within wind energy systems.