The development of deep learning provides a new research method for fault diagnosis. However, in the industrial field, the labeled samples are insufficient and the noise interference is strong so that raw data obtained by the sensor are occupied with noise signal. It is difficult to recognize time-domain fault signals under the severe noise environment. In order to solve these problems, the convolutional neural network (CNN) fusing frequency domain feature matching algorithm (FDFM), called CNN-FDFM, is proposed in this paper. FDFM extracts key frequency features from signals in the frequency domain, which can maintain high accuracy in the case of strong noise and limited samples. CNN automatically extracts features from time-domain signals, and by using dropout to simulate noise input and increasing the size of the first-layer convolutional kernel, the anti-noise ability of the network is improved. Softmax with temperature parameter T and D-S evidence theory are used to fuse the two models. As FDFM and CNN can provide different diagnostic information in frequency domain, and time domain, respectively, the fused model CNN-FDFM achieves higher accuracy under severe noise environment. In the experiment, when a signal-to-noise ratio (SNR) drops to -10 dB, the diagnosis accuracy of CNN-FDFM still reaches 93.33%, higher than CNN’s accuracy of 45.43%. Besides, when SNR is greater than -6 dB, the accuracy of CNN-FDFM is higher than 99%.

In order to prevent coal mine water inrush accidents, it is necessary to appropriately assess the water abundance of coal mines based on drilling and geophysical data. This paper studied a comprehensive risk assessment method of water inrush. First, a water inrush risk index was proposed based on the analytic hierarchy process-entropy method (AHP-EM) and the water-rich structure index was proposed based on the geological data coupled calculation, then weighted two indices above which established the comprehensive water inrush risk assessment method. Secondly, eight factors were chosen as risk control factors of water inrush: core recovery, aquifer thickness, distance from the indirect aquifer to the coal seam, aquiclude thickness, height of water-conducting fracture zone, sand-mud ratio, total layers of aquifer and aquiclude, and the equivalent thickness of sandstone. Finally, the No. 2 coal seam of Dahaize coal mine was taken as the research object, the factors were calculated, and a comprehensive water inrush assessment model was constructed. With site investigation and observation, the water inrush risk assessment model of the No.2 coal seam roof is consistent with the actual mining situation, which verifies the validity of the model. In addition, this method was used to evaluate the water-richness of the weathered bedrock fractured aquifer in the Zhangjiamao coal mine. The practical application of the two mines has verified the generality of the approach. The research could provide scientific assistance for mine water hazard mitigation and mining safety.

The calculation of the three-dimensional atmospheric dispersion model is often time-consuming, which makes the model difficult to apply to the emergency field. With the aim of addressing this problem, we propose a parallel computing algorithm for the CALPUFF atmospheric dispersion model. Existing methods for parallelizing the atmospheric dispersion model can be divided into two categories, with one using the parallel computing interface to rewrite the source code and the other directly dividing the repetitive elements in the computation task. This paper proposes an improved method based on the latter approach. Specifically, the method of spatial division with buffers is adopted to parallelize the wind field module of the CALPUFF model system, and the method for receptor layering is adopted to parallelize the dispersion module. In addition, the message queue software RabbitMQ is used as the communication middleware. A performance test is conducted on nine computing nodes on the Alibaba Cloud Computing Platform for a single-source continuous emergency leak case. The results show that the division method with a buffer of ten cells is most suitable for the case above in order to maintain the balance between computation speed and accuracy. This reduces the computation time of the model to about one-sixth, which is of great significance for extending the atmospheric dispersion model to the emergency field.