Xiaonan Feng

To identify the development stages of load-induced cracks in reinforced concrete beams under freeze–thaw and corrosion degradation, this paper proposes a mechanism-guided multi-branch neural network approach. Based on the physical meanings of acoustic emission parameters, the input features are divided into the following four groups: crack evolution process, damage intensity characterization, crack pattern characterization, and structural degradation state. Independent branches are employed for feature extraction, and hierarchical fusion is used to achieve collaborative representation of multidimensional damage information. Additionally, the Coyote Optimization Algorithm (COA) is introduced to optimize the initial parameters of key network layers. A total of 5789 valid acoustic emission event samples were obtained from loading tests, covering four degradation levels and four damage stages. Repeated experiments (20 trials) demonstrated that the proposed COA-MB-MLP model achieved an accuracy of 91.42% and a Macro-F1 score of 0.9118 on the test set, outperforming traditional benchmark models such as KNN, LDA, SVM, and RF, with statistically significant performance improvements. Ablation experiments revealed that the mechanism-guided multi-branch structure is the primary source of performance enhancement; COA initialization provides modest improvements in average test performance and reduces variability across repeated experiments. Feature response and confusion sample analysis indicated that amplitude is more sensitive to damage stage evolution, while features like counts, energy, and duration exhibit significant increases in the failure stage. Confusion between adjacent stages primarily stems from crack evolution continuity and ambiguous phase boundaries. The findings demonstrate that the proposed method effectively integrates multidimensional physical information from acoustic emission signals, providing a valuable reference for identifying crack development stages in freeze–thaw and corrosion-damaged reinforced concrete beams.

Raw mix-design variables used to predict ultra-high-performance concrete (UHPC) properties cannot fully represent internal proportions and structural compatibility. This study develops mechanism-informed proxy-core features based on particle packing, water film thickness, and rheology, and evaluates four feature systems using literature-derived datasets for compressive strength (924 samples), flexural strength (406 samples), and slump flow (192 samples). Eight regression models were compared using five-fold cross-validation, Bayesian optimization, SHAP, and ablation analysis. Proxy-W achieved an R² of 0.918 for compressive strength, only 0.005 higher than baseline-W on the original split. For flexural strength, baseline-WB achieved the highest R² (0.906), whereas proxy-W yielded the lowest MAE (2.140 MPa). The largest single-split difference occurred for slump flow (R²: 0.740 to 0.838 for W-based systems), but 30 repeated splits yielded mean R² values of 0.667 ± 0.169 and 0.738 ± 0.101 for baseline-W and proxy-W, respectively, while proxy-WB showed no average improvement over baseline-WB. Repeated and source-group validation further indicated that the predictive effects of proxy-core features were property-, representation-, partition-, and source-dependent. Overall, proxy-core features are best interpreted as physically informed relational representations rather than universally accuracy-enhancing features.

To investigate the evolution of acoustic emission (AE) parameters during the flexural failure of reinforced concrete (RC) beams subjected to freeze–thaw and corrosion, four RC beams were fabricated and assigned to four conditioning regimes: no deterioration, freeze–thaw only (75 cycles), corrosion only (4.8% mass loss), and combined. Three-point bending tests were conducted, combining AE and digital image correlation (DIC) techniques. The damage process was divided into four stages: micro-crack initiation, stable crack propagation, unstable crack propagation, and failure. The evolution of AE parameters including ring count, energy, amplitude, peak count, and duration was analyzed. Each parameter is positively correlated with load level and rises as the damage stage advances. The slope of cumulative parameters reflects crack development more reliably than instantaneous values. The effect of corrosion on these parameters is significantly greater than that of freeze–thaw. For corroded beams, AE parameter levels are higher during the micro-crack initiation stage but lower during the stable crack propagation stage. The overall AE activity decreases with increasing deterioration degree. High-amplitude events increase with damage progression, but fewer high-amplitude events are observed at the failure stage of severely deteriorated beams. This study reveals the correspondence between AE parameters and damage stages, providing an experimental basis for damage assessment using AE techniques.

Acoustic emission (AE) signals associated with different damage processes in reinforced concrete (RC) beams often show overlapping feature distributions, making unsupervised identification difficult. In this study, three RC beams were tested under loading-induced damage, freeze–thaw damage and reinforcement corrosion conditions, and 490 valid AE samples were obtained. Seven AE parameters were selected to construct the clustering feature set. To enhance the separation of different damage-related AE signals, a kernel K-means clustering framework was adopted, and the coyote optimization algorithm was used to optimize the kernel function type and key parameters based on the Gap Statistic. Comparative analysis with K-means, FCM and GMM was also conducted. The results show that the COA-optimized kernel K-means method achieved the best overall clustering performance, increasing the mean ACC by 10.35 percentage points compared with K-means and by 2.49 percentage points compared with GMM. Its mean ARI was also higher than that of GMM by 0.0330, while the standard deviation of ACC decreased from 7.73% to 4.19%. Class-level results indicated that loading-induced AE signals were more readily identified, whereas freeze–thaw and corrosion signals were more affected by feature overlap. Feature interpretation further showed that the main misclassified samples were located in transitional feature regions. The results suggest that COA-based kernel optimization can improve the clustering separation and stability of AE signal identification for different damage types in RC beams.

This study investigated the effects of varying water–binder (w/b) ratios and internal curing materials—superabsorbent polymer (SAP) and ceramsite—on the shrinkage behavior and crack resistance of ultra-high-performance concrete (UHPC). Although internal curing has been extensively studied, the comparative effectiveness of different internal curing materials on early-age shrinkage and restrained cracking behavior of UHPC under consistent mixture proportions remains unclear. To address this gap, a systematic experimental comparison of SAP and ceramsite was conducted. The influences of w/b ratio and different amounts and addition methods (dry and pre-absorbed addition) of SAP and ceramsite on the flowability, mechanical properties, early autogenous shrinkage, drying shrinkage, and early crack resistance of UHPC were discussed. Findings indicate that increasing the w/b ratio reduces autogenous shrinkage but compromises mechanical properties, altering the cracking mode from primary microcracks to a few wider cracks. Pre-saturated ceramsite (less than 10% volume) and SAP effectively mitigate autogenous and drying shrinkage, enhancing crack resistance without significantly reducing mechanical properties. However, exceeding a ceramsite volume dosage of 10% or using the dry addition method increased the flowability of UHPC, while decreasing crack resistance. Microstructural analysis reveals that internal curing materials facilitate hydration and enhance structural density through the formation of ettringite and calcium silicate hydrate. To optimize shrinkage reduction while maintaining mechanical properties, SAP should be incorporated in a dry state, with a dosage limited to 0.4% of the mass of the cementitious material; ceramsite needs to be pre-saturated and limited to 5% of the total volume.