The PCHE, as an efficient heat exchanger, plays a crucial role in the storage and regasification of LNG. However, among the existing studies, those that integrate this field with deep learning are scarce. Moreover, research on explainability remains insufficient. To address these gaps, this study first constructs a dataset of heat transfer coefficients (h) through numerical simulations. Pearson correlation analysis is employed to screen out the most influential features. In terms of predictive modeling, the study compares five traditional machine learning models alongside deep learning models such as long short-term memory neural networks (LSTMs), gated recurrent units (GRUs), and Transformer. To further enhance prediction accuracy, three attention mechanisms—self-attention mechanism (SA), squeeze-and-excitation mechanism (SE), and local attention mechanism (LA)—are incorporated into the deep learning models. The experimental results demonstrate that the artificial neural network achieves the best performance among the traditional models, with a prediction accuracy for straight-path h reaching 0.891799 (R²). When comparing deep learning models augmented with attention mechanisms against the baseline models, both LSTM–SE in the linear flow channel and Transformer–LA in the hexagonal flow channel exhibit improved prediction accuracy. Notably, in predicting the heat transfer coefficient of the hexagonal channel, the determination coefficient (R²) of the Transformer–LA model reaches 0.9993, indicating excellent prediction performance. Additionally, this study introduces the SHAP interpretable analysis method to elucidate model predictions, revealing the contributions of different features to model outputs. For instance, in a straight flow channel, the hydraulic diameter (Dh) contributes most significantly to the model output, whereas in a hexagonal flow channel, wall temperature (Tinw) and heat flux (Qw) play more prominent roles. In conclusion, this study offers novel insights and methodologies for PCHE performance prediction by leveraging various machine learning and deep learning models enhanced with attention mechanisms and incorporating explainable analysis methods. These findings not only validate the efficacy of machine learning and deep learning in complex heat exchanger modeling but also provide critical theoretical support for engineering optimization.

Traditional methods for picking small-target crops like pepper are time-consuming, labor-intensive, and costly, whereas deep learning-based object detection algorithms can rapidly identify mature peppers and guide mechanical arms for automated picking. Aiming at the low detection accuracy of peppers in natural field environments (due to small target size and complex backgrounds), this study proposes an improved Yolov8n-based algorithm (named Yolov8n-RCP, where RCP stands for RVB-CA-Pepper) for accurate mature pepper detection. The acronym directly reflects the algorithm’s core design: integrating the Reverse Bottleneck (RVB) module for lightweight feature extraction and the Coordinate Attention (CA) mechanism for background noise suppression, dedicated to mature pepper detection in complex crop environments. Three key optimizations are implemented: (1) The proposed C2F_RVB module enhances the model’s comprehension of input positional structure while maintaining the same parameter count (3.46 M) as the baseline. By fusing RepViTBlocks (for structural reparameterization) and EMA multi-scale attention (for color feature optimization), it improves feature extraction efficiency—specifically, reducing small target-related redundant FLOPs by 18% and achieving a small-pepper edge IoU of 92% (evaluated via standard edge matching with ground-truth annotations)—thus avoiding the precision-complexity trade-off. (2) The feature extraction network is optimized to retain a lightweight architecture (suitable for real-time deployment) while boosting precision. (3) The Coordinate Attention (CA) mechanism is integrated into the feature extraction network to suppress low-level feature noise. Experimental results show that Yolov8n-RCP achieves 96.4% precision (P), 91.1% recall (R), 96.2% mAP0.5, 84.7% mAP0.5:0.95, and 90.74 FPS—representing increases of 3.5%, 6.1%, 4.4%, 8.1%, and 11.58FPS, respectively, compared to the Yolov8n baseline. With high detection precision and fast recognition speed, this method enables accurate mature pepper detection in natural environments, thereby providing technical support for electrically driven automated pepper-picking systems—a critical application scenario in agricultural electrification.

Fractional-order time-delay systems boast better dynamic performance than integer-order ones in optimally controlling industrial design objects. However, in lack of commendable methodologies, designing proper controllers for these systems confronts a plurality of challenges. This study puts forth an innovative design approach that merges frequency-domain analysis with time-domain optimization concepts, so that fractional-order Tilt-Integral-Derivative (TID) controllers can be acquired. To pursue a stable control system loop, the tilted and integral gains of fractional-order TID controllers are identified as per frequency-domain specifications, including gain crossover frequency and phase margin. In light of these specifications (e.g., the integral of time-weighted absolute error (ITAE)), the differential gain and fractional-order operator $\lambda$ of the controller are determined, which accomplishes a desirable dynamic performance in the time domain. This article expounds on the procedure of how to develop the proposed fractional-order TID controller and furnishes illustrative examples for the research steps. As manifested by the simulation results, the proposed controller dramatically upgrades the control performance of the system in contrast to conventional PID, FOPI, and FOPID controllers. Moreover, it outperforms PID and fuzzy PID in terms of responding to the demand variations in step signals.

Lithium-ion batteries exhibit pronounced nonlinear, time-varying, and hysteretic behavior under complex dynamic operating conditions, and their state of charge (SOC) is jointly influenced by the current operating state and the preceding charge-discharge history. To exploit these complementary capabilities, a Transformer-BiLSTM model is developed for lithium-ion battery SOC estimation. Current and terminal voltage are used as inputs, and historical sequence samples are constructed using a sliding window of length 50. The raw features are first processed by linear projection and sinusoidal positional encoding and then fed into a two-layer Transformer encoder, in which multi-head self-attention captures global dependencies across sampling instants. The encoded features are subsequently passed through a two-layer stacked BiLSTM to further characterize local temporal dynamics associated with polarization, diffusion, and dynamic load variations. A dropout layer and a fully connected layer are finally used to obtain the SOC estimate. The model is jointly trained on multiple operating cycles and evaluated on unseen NEDC and UDDS cycles to assess its generalization across driving cycles. Experimental results show that the proposed model accurately tracks SOC variations under different dynamic conditions. For the NEDC and UDDS cycles, the RMSE values are 1.57% and 1.07%, the MAE values are 1.10% and 0.76%, and the coefficients of determination (R²) are 0.9953 and 0.9978, respectively. Compared with the Transformer, BiGRU, BiLSTM, and Transformer-BiGRU models, the proposed method achieves superior overall performance in estimation accuracy, error stability, and generalization across driving cycles.