The mechanical characteristics and mechanisms of rock failure involve complex rock mass mechanics problems involving parameters such as energy concentration, storage, dissipation, and release. Therefore, it is important to select appropriate monitoring technologies to carry out relevant research. Fortunately, infrared thermal imaging monitoring technology has obvious advantages in the experimental study of rock failure processes and energy dissipation and release characteristics under load damage. Therefore, it is necessary to establish the theoretical relationship between the strain energy and infrared radiation information of sandstone and to reveal its fracture energy dissipation and disaster mechanism. In this study, an MTS electro-hydraulic servo press was used to carry out uniaxial loading experiments on sandstone. The characteristics of dissipated energy, elastic energy, and infrared radiation during the damage process of sandstone were studied using infrared thermal imaging technology. The results show that (1) the transition of sandstone loading from one stable state to another occurs in the form of an abrupt change. This sudden change is characterized by the simultaneous occurrence of elastic energy release, dissipative energy surging, and infrared radiation count (IRC) surging, and it has the characteristics of a short duration and large amplitude variation. (2) With the increase in the elastic energy variation, the surge in the IRC of sandstone samples presents three different development stages, namely fluctuation (stage Ⅰ), steady rise (stage Ⅱ), and rapid rise (stage Ⅲ). (3) The more obvious the surge in the IRC, the greater the degree of local damage of the sandstone and the greater the range of the corresponding elastic energy change (or dissipation energy change). (4) A method of sandstone microcrack location and propagation pattern recognition based on infrared thermal imaging technology is proposed. This method can dynamically generate the distribution nephograph of tension-shear microcracks of the bearing rock and accurately evaluate the real-time process of rock damage evolution. Finally, this study can provide a theoretical basis for rock stability, safety monitoring, and early warning.

It is urgent to improve the comprehensive utilization rate of tailings, especially to solve the problem of fine iron ore tailings treatment and the shortage of natural sand, so iron ore tailings concrete with different replace rates was prepared in this paper. At the same time, in order to further popularize the wide application of iron tailings concrete, it is of great significance to study the frost resistance and life prediction of it in cold regions. Thereby, a series of experimental research on iron ore tailings concrete under freezing-thawing cycles was carried out to analyze the attenuation law of quality, compressive strength, and relative dynamic elastic modulus after different freezing-thawing cycles. In this paper, the relative dynamic elastic modulus was used to define the damage factor, and the freezing-thawing damage model of iron ore tailings concrete was established based on Weibull distribution. Here, the specific function form of two-parameter Weibull distribution of iron ore tailings concrete under freezing-thawing cycle was proposed, and the model was used to predict life of it. The results showed that, compared with the replacement rates of 20%, 60%, and 80%, the iron ore tailings with 40% replacement rate had the best improvement effect on the frost resistance of concrete and can delay the attenuation of the compressive strength of concrete. This model can not only predict the durability life of iron ore tailings concrete under freezing-thawing conditions, but also provide theoretical reference for the application of iron ore tailings concrete in cold areas.

In order to predict the relative dynamic elastic modulus (RDEM), which is used to reflect the frost resistance of iron ore tailings concrete, the backpropagation neural network (BPNN) was used in this study. Here, one hidden layer was chosen in the structure of BPNN. It is well known that the number of neurons in the hidden layer is the key of BPNN; hence, checking the features of overfitting was chosen in this paper to determine the number of neurons in the hidden layer. According to the actual conditions of freeze–thaw cycle test, a BPNN model of 2D input vector and 1D output vector was established. Thirty datasets from the test were used for training and test the proposed ANN. The results showed that the predicted values were in good agreement with the measured ones, and the correlation coefficient between them reached 0.9505. It showed that the BPNN, with the ability to solve nonlinear problems, had great advantages in predicting RDEM, and it can be used as reliable and simple forecasting tools for the prediction of frost resistance of iron ore tailings concrete.

The incorporation of waste rubber in concrete is a beneficial solution, it not only can improve the frost-resistance of concrete, reduce environmental pollution, and it also has a role that cannot be ignored in reducing carbon emissions. Benefit from the relative dynamic elastic modulus, the frost-resistance of concrete can be effectively evaluated. Hence, predicting the frost resistance accurately of rubber concrete is urgently needed for its popularization and application in cold areas. In this paper, 153 sets of relative dynamic elastic modulus data of rubber concrete were collected from published literatures, by combining information entropy and back propagation neural network (BPNN), a two-stage hybrid BPNN (5-7-1) prediction model was proposed. In this model, information entropy was used for input features selection, and BPNN was used for predicting frost resistance. The number of neurons in the hidden layer of BPNN was comprehensively determined by empirical formula and the minimum mean square error of the model. The prediction results show that after features selection, the MSE (Mean Square Error), RMSE (Root Mean Square Error), MAE (Mean absolute error), MAPE (average absolute percentage error) and R ² (determination coefficient) are all superior to the BPNN without feature selection. The results of this research provide a new idea for the frost-resistance prediction for rubberized concrete.