Aiming at the problem of insufficient accuracy caused by the insufficient mining of spatiotemporal features in the process of unsafe behavior and danger identification of construction personnel, the traditional two-stream convolution model is improved, and a two-stream convolution dangerous behavior recognition model based on Faster R-CNN-LSTM is proposed. In this model, the Faster R-CNN network is connected in parallel with the LSTM network. The Faster R-CNN network is used as the spatial flow, and the human spatial motion posture is divided into static and dynamic features to extract the anchor point features, respectively. The fusion of the two is used as the output of the spatial flow. An improved sliding long-term and short-term memory network is used in the time flow to increase the extraction ability of the time series features of the construction personnel. Finally, the two branches are fused in time and space to classify and identify whether the construction personnel wear safety helmets. The results show that the MAP of the improved Faster R-CNN-LSTM network framework is increased by 15%. The original CNN-LSTM network framework detected four targets, but there was one misdetection, with an accuracy of 91.48%. The improved frame detection accuracy reaches 99.99%, and there is no error detection. The proposed method is superior to the pre-improvement and other methods that can effectively identify the unsafe behavior of construction workers on construction sites and also has a good distinction effect on fuzzy actions.

Compared to continuous abrasive jet and high-pressure water jet, pulsed abrasive water jet can intermittently generate high water hammer pressure and is commonly used in applications such as coal mining and petroleum engineering for rock-breaking purposes. Pulse frequency is one of the key factors affecting the evolution of abrasive acceleration and jet impact energy, thereby influencing the system's energy transfer efficiency and cutting ability. The equation of motion for abrasive acceleration is established based on the two-phase flow theory, and the variation law of abrasive velocity with pulse frequency is solved. Fluent-EDEM (Extended Discrete Element Method) was used to simulate the abrasive water jet flow field structure and abrasive acceleration process under different pulse frequency conditions, and the intrinsic influence mechanism of pulse frequency on impact energy was clarified. Combined with rock-breaking experiments using pulsed abrasive water jets, the following conclusions are drawn: as the pulse frequency increases, the inertia of the abrasive makes it unable to respond in a timely manner to the high pulsation frequency of water, resulting in a reduction in the exchange rate of intermittent energy and a gradual decrease in the maximum velocity of the abrasive. This trend is more pronounced at a higher pulse frequency. Within the same time period, the cumulative impact energy of the pulsed abrasive water jet decreases first and then increases with the increase in pulse frequency, indicating the existence of an optimal impact frequency that maximizes the cumulative impact energy of the pulsed abrasive water jet. Under fixed jet pressure conditions and the same erosion time, the pulsed abrasive water jet with a frequency of 30 Hz exhibits the best rock-breaking capability.

Soft coal is characterized by low strength and weak bonds, which play a key role in the occurrence and development of dynamic disasters. A more thorough understanding of the failure mechanics and infrared radiation characteristics of soft coal at various moisture contents is needed. In this study, infrared radiation experiments were conducted for soft coal at various moisture contents. The results indicate that moisture content affects compressive strength and elastic modulus of soft coal with compressive strength and elastic modulus being highest at moderate. Moisture has a substantial influence on the infrared radiation of compressed coal samples. Soft coals with high moisture contents show a smaller fluctuation of the average infrared radiation temperature curve and a stronger direct relationship of temperature and stress in the linear elastic stage. The coal samples with various moisture contents show a warming trend during loading. The increased amplitude of infrared radiation temperature per unit stress shows a linear relationship with moisture content. A large number of high temperature red spots appear along the diagonal line of coal samples with 0% moisture content before fracture, while changes in the infrared thermal image of the coal sample with 4% moisture content were negligible during loading.

To further improve the accuracy of recurrent neural network in predicting the gas concentration in the upper corner of the mine tunnel, this paper proposes a method to construct a gas concentration prediction model based on multiple sequence long and short memory network, considering the spatial correlation between the gas concentration in the return airway and upper corner. The reliability of the model construction is improved by using the white noise test and smoothness test to verify the interpretability of the data in this paper and constructing supervised learning type data for gas concentration prediction model training and testing by means of data set division and data windowing. Through experimental comparison, grid search, and time series decomposition, the model algorithm, training parameters, and experimental results were combined to make an in-depth analysis of the influence of each parameter on the model training and the prediction. A training model of the spatially fused gas concentration prediction model with a network layer of 1 and a number of neurons of 32 as the model structure, Adam as the optimization algorithm, and a learning rate of 0.001 and a batch size of 32 as the training parameters was finally determined. The gas concentration prediction model trained in this paper performed well in the test set with a mean square error (MSE) of 0.0013, and its superiority was verified by comparing it with other models to provide some experience and basis for subsequent studies on gas concentration prediction in the upper corner.

With the continuous increase in coal mining depth, gas-related disasters have become a critical constraint for the safe and efficient extraction of coal resources. To address prominent issues such as the lack of scientific evaluation in extraction design, insufficient dynamic monitoring during the process, and delayed effectiveness assessment, this study establishes a full-cycle evaluation framework for gas extraction, encompassing the entire process from design and construction to implementation. A Bayesian optimization-based Random Forest Regression model (BO-RFR) is proposed for the preliminary evaluation of extraction schemes, while a reduced-order prediction model integrating deep neural networks and convolutional autoencoders (DNN-CAE) is developed to enable dynamic prediction and reconstruction of residual gas pressure fields. Field tests demonstrated that the BO-RFR model achieved a residual gas pressure prediction error of less than 0.02 MPa, while the DNN-CAE dynamic evaluation model yielded an average mean squared error (MSE) of 2.73 × 10^–5 and mean absolute error (MAE) of 0.00493, confirming the high accuracy and reliability of the proposed method in engineering applications. The results of this study can provide scientific decision-making support for coal mine gas control, enabling precise control and intelligent management of the extraction process. This holds significant practical value for enhancing the safety of coal mine production.

[This corrects the article DOI: 10.1021/acsomega.2c05188.].