Improved DenseNet-ECA Network with Continuous Wavelet Transform for High-Accuracy Rolling Bearing Fault Diagnosis

Stable operation of rolling bearings is crucial in mechanical systems, as failures can lead to severe performance degradation and shutdowns. Early fault detection significantly reduces maintenance costs and safety risks, yet traditional diagnosis methods often struggle with inefficiency and poor adaptability. This paper presents an improved DenseNet-ECA network combined with Continuous Wavelet Transform (CWT) for signal preprocessing to extract key features. The model achieves 100% fault detection accuracy across various speeds, showcasing superior performance and stability under diverse conditions, providing a promising solution for rolling bearing fault detection.