The failure of strata with different strength under coal pillar in close distance coal seams has its particularity. Theoretical analysis, numerical simulation, and similar test were used to obtain the failure mechanism of the soft and hard stratification of floor strata under the boundary of wide coal pillar and make out the influence of the strata failure on roadway layout. It shows that the failure of the soft and hard stratification is not synchronous and with different failure forms. The deformation and plastic failure of the soft strata occurs at first; then the fracture of the main bearing hard stratum occurs. Based on the main bearing stratum unbalanced forces acted by upper and lower strata, the initial fracture position and broken blocks span were calculated by bending fracture mechanics model of ultimate span beam. The coal pillar side of the roadway under coal pillar forms “column-beam-column” superimposed structure. The roadway layout under coal pillar and under coal pillar boundary with close distance has large influence on the plastic extension of the “column-beam-column” superimposed structure and the broken of main bearing stratum. The influence of roadway arrangement in floor strata outside coal pillar is smaller than those under coal pillar. It is difficult to control the deformation of roadway surrounding rock if the roadway locates beneath the broken blocks of the main bearing stratum.

Multiple roadways are opened to solve the difficulties in gas control in the high seam working face of high gassy mines, resulting in a large width of coal pillars between roadway groups. Moreover, recovering coal pillars is challenging during stoping, which causes serious resource waste. While the size of coal pillars is optimized, the reasonable horizon layout of roadway groups is performed to ensure a fine recovery rate of coal resources, mine safety, and efficient production. Theoretical analysis was applied to calculate the stress distribution of surrounding rocks around coal pillars based on the engineering practice of the W2302 working face in the Sihe Coal Mine. The MathCAD software Prime 2.0 combined with the on-site measured stress was used to draw the stress distribution under different coal pillar widths, which can derive the stress distribution data of the floor. The vertical stress concentration factors tended to be consistent at 10 m below the coal pillars and about 12 m from the centerline of coal pillars under different coal pillar widths, this area was suitable for the layout of floor rock roadways with a stress concentration factor of about 1. Numerical simulations were used to compare the stress distributions of the roadway floor and the deformation features of surrounding rocks under different coal pillar widths. The coal pillar width was finally determined to be 45 m, which optimized the size of coal pillars. The return air roadways were arranged at 9.8 m below the coal seam combined with the strata distribution on-site. According to the industrial test on-site, the return air roadways were available during the service period, which showed reliable theoretical analysis and simulation results. A reasonable horizon layout of return air roadways can provide a scientific basis and reference for similar projects.

In order to reasonably evaluate the risk of rock bursts in mines, the factors impacting rock bursts and the existing grading criterion on the risk of rock bursts were studied. By building a model of hierarchic analysis method, the natural factors, technology factors, and management factors that influence rock bursts were analyzed and researched, which determined the degree of each factor’s influence (i.e., weight) and comprehensive index. Then the grade of rock burst risk was assessed. The results showed that the assessment level generated by the model accurately reflected the actual risk degree of rock bursts in mines. The model improved the maneuverability and practicability of existing evaluation criteria and also enhanced the accuracy and science of rock burst risk assessment.

To address the problem of the concrete filling body (CFB) force failing to reach test strength in remaining roadways, the weakening effects due to aspect ratio and dimensional parameters on the actual CFB strength were investigated in this study. The geometric effects of CFB (including hoop and size effects) as well as the geometric effect coefficient determination method were analysed. Through laboratory tests and PFC numerical simulations, the hoop and size effect coefficients of the CFB in the Gaohe Coal Mine were studied. Furthermore, the calculation equations of actual strength and bearing capacity of the CFB were derived. Regarding the filling body failure and coal deformation in the remaining roadway located at the No. W1319 working face, the actual bearing capacity of CFB and surrounding rock stability during secondary exploitation were theoretically studied. The investigation suggests the adoption of a grouting reinforcement scheme for surrounding rock. The field applications performed have demonstrated that the deformation control effect in the remaining-roadway surrounding rock was effectively improved during second mining and the filling body beside the roadway suffered no additional damage. Studying the geometric effect of CFB can provide some theoretical guidance and industrial significance to accurately identify the filling body strength and reduce the failure risk of surrounding rock in remaining roadways.

Image classification techniques have succeeded greatly on various large-scale visual datasets using deep convolution neural networks. However, previous deep models usually suffer severe performance degradation in highly skewed datasets, which restricts their practical application. In this paper, we propose a novel Hierarchical Rebalancing Dual-Classifier model for long-tailed recognition. To better identify the tail samples and maintain the performance of head classes, we propose a dual-classifier framework with a uniform sampler for performing their duties. For balancing the learning of feature representation and classifiers, a dynamic weight is introduced to adjust the model’s attention. To alleviate the feature deviation between training data and testing data, a hierarchical rebalancing loss is designed for the re-weighting branch, which adjusts the decision values in predicted logits to facilitate the model actively compensating for tail categories. Finally, we conduct extensive experiments on standard long-tailed benchmarks Cifar10-LT, Cifar100-LT, ImageNet-LT, and iNaturalist2018, demonstrating the effectiveness and superiority of our HRDC.

The preevaluation to the vulnerability of the surrounding rocks is proposed as one of the reliable indicators of the safety coefficient of gateway support and a foundation to optimize the support design parameters. In this study, taking the surrounding rocks, stress, geological environment, and service time into consideration, the safety coefficient is determined based on the vulnerability scores calculated by the vulnerability preevaluation model of the surrounding rock. Applying the safety coefficient to the instability evaluation of the composite rock‐bolt bearing structure, the strength required to maintain the stability of the gateway is calculated, which further provides references and guidance to the optimization of the anchor support parameters. This method has been successfully adopted by the GuCheng coal mining project in N1303 tailgate to strengthen the anchor‐bolt structure in the roof watering area especially the main inclined shaft. Applying more accurately calculated strength to the anchor‐bolt structure can effectively avoid the issue of overcompensation, thus reducing the cost and increasing the driving speed. Furthermore, this method provides insights into optimizing design parameters of the gateway. This method provides a reliable basis for the optimization design of bolt support parameters in coal mine gateway.