In many steel mills, the working life of continuous casting machine rollers is relatively short, requiring frequent replacement and negatively impacting production efficiency. To effectively extend the service life of these rollers, this study focuses on a continuous casting machine at a steel mill in China. A numerical simulation was conducted, revealing that the rollers in contact with the high-temperature casting billet experience significant thermal and stress impacts. The traditional cooling channel struggles to sufficiently reduce both the surface temperature and stress, resulting in severe thermal fatigue damage to the roller surfaces. Observations of roller surface wear showed signs of adhesive wear, fatigue cracks, and spalling occurring in various regions of the roller, which aligned with the stress distribution predicted by the simulation. In response, the cooling channel structure was modified to enhance the cooling effect of the water. Optimization of both the cooling channel structure and its parameters was carried out using a coupled flow-heat-force numerical simulation method. The optimized cooling channel effectively improved the working condition of the continuous casting roll, as the maximum temperature of the roll surface was reduced from 810 K to 591 K, the circumferential temperature difference was reduced by 38%, and the maximum equivalent stress decreased from 791 MPa to 558 MPa. This adjustment also resulted in a more uniform surface temperature distribution, mitigating the sudden fluctuations in normal stress that are typical of conventional rollers.

Steel reinforcement in marine concrete structures is vulnerable to chloride-induced corrosion, which compromises its structural integrity and durability. This study explores the combined effect of the alloying element Cr and the smart corrosion inhibitor LDH-NO₂ on enhancing the corrosion resistance of steel reinforcement. Employing a machine learning approach with a support vector machine (SVM) algorithm, a predictive model was developed to estimate the polarization resistance of steel, considering Cr content, LDH-NO₂ dosage, environmental pH, and chloride concentration. The model was rigorously trained and validated, demonstrating high accuracy, with a correlation coefficient exceeding 0.85. The findings reveal that the addition of Cr and application of LDH-NO₂ synergistically improve corrosion resistance, with the model providing actionable insights for selecting effective corrosion protection methods in diverse concrete environments.

The number of grains per sea rice panicle is an important parameter directly related to rice yield, and it is also a very important agronomic trait in research related to sea rice breeding. However, the grain number per sea rice panicle still mainly relies on manual calculation, which has the disadvantages of being time-consuming, error-prone, and labor-intensive. In this study, a novel method was developed for the automatic calculation of the grain number per rice panicle based on a deep convolutional neural network. Firstly, some sea rice panicle images were collected in complex field environment and annotated to establish the sea rice panicle image data set. Then, a sea grain detection model was developed using the Faster R-CNN embedded with a feature pyramid network (FPN) for grain identification and location. Also, ROI Align was used to replace ROI pooling to solve the problem of relatively large deviations in the prediction frame when the model detected small grains. Finally, the mAP (mean Average Precision) and accuracy of the sea grain detection model were 90.1% and 94.9%, demonstrating that the proposed method had high accuracy in identifying and locating sea grains. The sea rice grain detection model can quickly and accurately predict the number of grains per panicle, providing an effective, convenient, and low-cost tool for yield evaluation, crop breeding, and genetic research. It also has great potential in assisting phenotypic research.

To balance machining efficiency and surface quality in the electrical discharge machining (EDM) of 304 stainless steel, this study investigates the effects of peak current, pulse duration, and pulse interval on material removal rate (MRR) and surface roughness (Ra). Orthogonal experiments, optimal Latin hypercube sampling, grey relational analysis, and single-factor experiments were employed to characterize the EDM process and clarify the effects of electrical parameters on surface morphology and roughness. The results indicate that peak current has the most significant influence on both MRR and Ra, while changes in electrical parameters affect Ra through discharge intensity, thermal input, and discharge crater characteristics. A hybrid RBF-Kriging surrogate model was then developed to describe the nonlinear relationship between process parameters and responses, achieving R² values of 0.9773 for MRR and 0.9874 for Ra. Under a unified repeated 5-fold cross-validation and LOOCV protocol, this model performed comparably to quadratic RSM, GPR, SVR and ANN (all R² > 0.95). NSGA-II was subsequently used to obtain the Pareto-optimal solutions, and equal-weight TOPSIS was applied to determine a compromise solution. The optimal parameters were a peak current of 5.78 A, pulse duration of 47 μs, and pulse interval of 22 μs, yielding predicted MRR and Ra values of 3.338 mm³/min and 3.936 μm, respectively. Experimental verification produced MRR and Ra values of 3.566 mm³/min and 4.054 μm, with prediction errors of 6.81% and 2.98%, respectively. The results demonstrate that the proposed RBF-Kriging-NSGA-II framework provides an effective approach for balancing machining efficiency and surface quality in EDM of 304 stainless steel.