Intelligent bearing fault diagnosis based on deep learning has immense potential. However, improving the noise immunity, generality, and accuracy of fault diagnosis methods is still challenging. This paper proposes a novel adaptive hybrid attention mechanism deep residual threshold network (AHA-RTN) for bearing fault diagnosis under various noise conditions. First, channel-wise and spatial attention were both integrated into residual blocks to capture multiscale information. Hybrid attention was obtained using the proposed adaptive attention module, which computes adjustment coefficients of each attentional mechanism. Next, a novel noise reduction activation function based on soft thresholding was incorporated to suppress noise. Finally, the method was validated on two distinct bearing datasets under various noise conditions. The results show that the proposed AHA-RTN has better noise immunity and accuracy than the other advanced multiscale convolutional neural networks.

Stratospheric airships are long-endurance aerostats and have broad applications. All of the energy required for their operation is obtained from solar radiation, which makes accurate calculation of the energy output from the solar array crucial to the design and flight planning of the airships. However, the status of each photovoltaic module in the solar array may differ due to the airship curvature, resulting in mismatch losses and lowered output power, which has not been widely studied. In this paper, an irradiation model and a thermal model are established based on the actual arrangement of the modules. The output power model is established considering the non-uniform radiation in the array. The mismatch losses of the array are analyzed under different flight conditions. The output power of the solar array is decreased by up to 31.6% compared to the ideal state. Moreover, the proportion of mismatch losses increases with latitude, but the maximum mismatch loss power occurs at mid-latitudes. Then, an array reconfiguration method is proposed based on the irradiance dispersion index and position dispersion index. The reconfigured array increases output power by 11.5% and can maintain energy balance in continuous flight. The results can be used to correct the overestimation of the output power during the airship design or to guide the configuration of the solar array.