Attention-Based Deep Learning Method for Rotor Temperature Estimation in Permanent Magnet Synchronous Motors
Hao Jing, Zifeng Chen, Lefei Ge, Xueqing Wang, Dianxun Xiao
The Hong Kong University of Science and Technology (Guangzhou) Northwestern Polytechnical University Sichuan University
阅读操作
确认中在文库中上传 PDF 后可生成中文音频讲解。
摘要与影响
Accurate real-time temperature estimation in permanent magnet synchronous motors (PMSMs) is critical for safe and efficient operation. This paper presents an attentionbased deep learning framework that addresses key challenges of practical deployment: model efficiency, stability, and ease of implementation. The proposed lightweight encoder-GRU architecture is systematically evaluated against conventional deep learning benchmarks and a full encoder-decoder model. Experimental results demonstrate that the proposed model achieves a competitive mean absolute error (MAE) of 1.67°C with a parameter count of only approximately 500, a fraction of that required by other deep learning methods. Crucially, the architecture exhibits superior stability, significantly reducing the maximum prediction error and demonstrating high reliability during complex operational transients. Furthermore, the framework's versatility is showcased by its successful extension into a multi-task learning (MTL) model, which employs a taskgrouped decoding strategy to provide a holistic and accurate thermal profile of both the rotor and stator. By delivering a solution that excels in parameter efficiency and robustness without requiring prior physical modeling, this work validates a practical and accessible pathway for deploying advanced, realtime thermal monitoring on resource-constrained embedded systems.
逐年被引趋势
关键指标
同类平均 = 1
同领域 · 同年份 · 同类型
Google Scholar 与 OpenAlex 的被引统计范围不同,数值存在差异属正常。
AI 辅助阅读
依据:摘要
可就本文提问;依据不足时会说明。
学术脉络
学科主题
工程Electric Motor Design and Analysis
Sensorless Control of Electric Motors · Machine Fault Diagnosis Techniques