Research on Intelligent Diagnosis for Equipment Fault of Rotary Machinery Based on Adaptive Wavelet Convolutional Capsule Network

By taking rotary machine as the research object, a fault detection method based on improved capsule network and adaptive wavelet noise reduction is proposed to guarantee the stability of daily operation of mechanical equipment. Among them, capsule network is used as the basic fault detection method, which is improved by introducing residual module and other methods. In addition, the fault detection performance is further improved by combining the method of adaptive wavelet noise reduction. The experimental results show that after introducing adaptive wavelet noise reduction method, the detection accuracy of the constructed detection method in noisy environments is significantly improves, indicating that the introduction of adaptive wavelet noise reduction is necessary. Compared with other commonly used detection methods, the fault detection method based on improved capsule network and adaptive wavelet noise reduction has better fault detection performance, and the detection accuracy reaches 99.89%. The fluctuation range is 99.89% ± 0.15%, with good stability. Meanwhile, in the noisy environment, the detection accuracy of the method fluctuates less, indicating that it has better anti-noise ability. In summary, the equipment fault detection method based on improved capsule network and adaptive wavelet noise reduction has excellent performance and good anti-noise ability, and it can be applied to the actual working scene of rotary machine for fault detection, effectively ensuring the normal operation of the equipment.