No abstract

With the increasingly widespread application of deep-learning technology in the field of coal mines, the image recognition of mine water inrush has become a hot research topic. Underground environments are complex, and images have high noise and low brightness. Additionally mine water inrush is accidental, and few actual image samples are available. Therefore, this paper proposes an algorithm that recognizes mine water inrush images based on few-shot deep learning. According to the characteristics of images with coal wall water seepage, A bilinear neural network was used to extract the image features and enhance the network's fine-grained image recognition. First, features was extracted using a bilinear convolutional neural network. Second, the network was pre-trained based on cosine similarity. Finally, the network was fine-tuned on the predicted image. For single-line feature extraction, the method is compared with big data and few-shot learning. According to the experimental results, the recognition rate reaches 95.2% for few-shot learning based on bilinear neural network, thus demonstrating its effectiveness.

Disastrous water inrush often causes heavy property losses and even casualties, while the current theory and technology of mine water prevention and control can not accurately predict the time and place of disastrous water inrush in mine, so monitoring disastrous water inrush has become an urgent problem to be solved. Through the networked deployment of water level sensors in the roadway, this paper proposes a monitoring network for catastrophic water inrush in coal mine, and studies a multi-constraint and multi-objective optimal deployment method for the monitoring network. By setting three practical constraints of mining area, risk level of water inrush and installation at specified location, and constructing two objective functions of minimum total cost and minimum average monitoring time, a mathematical model is established, and then NSGA - Ⅱ multi-objective genetic algorithm is designed to solve the model. As results, the method can optimize the monitoring network for mine water inrush from two dimensions of time and space. The proposed method is then verified in the Beiyangzhuang coal mine in the North China. The results show that the average time of the catastrophic water inrush monitored can be controlled within 916 seconds by using only 28 water level sensors, and the higher the risk level of water inrush, the shorter the monitoring response time. Due to the constraints of installation at specified location, the monitoring network can also take into account monitoring the daily water inflow in the Beiyangzhuang coal mine.