This study focuses on the problem of corn yield prediction, and a novel prediction model based on a dynamic ensemble stacking regression algorithm is proposed. The model aims to achieve more accurate corn yield prediction based on the in-depth exploration of the potential correlations in multisource and multidimensional data. Data on the weather conditions, mechanization degree, and maize yield in Qiqihar City, Heilongjiang Province, from 1995 to 2022, are used. Important features are determined and extracted effectively by using principal component analysis and indicator contribution assessment methods. Based on the combination of an early stopping mechanism and parameter grid search optimization, the performance of eight base models, including a deep learning model, is fine-tuned. Based on the theory of heterogeneous ensemble learning, a threshold is established to stack the high-performing models, realizing a dynamic ensemble mechanism and employing averaging and optimized weighting methods for prediction. The results demonstrate that the prediction accuracy of the proposed dynamic ensemble regression model is significantly better as compared to the individual base models, with the mean squared error (MSE) being as low as 0.006, the root mean squared error (RMSE) being 0.077, the mean absolute error (MAE) being 0.061, and a high coefficient of determination value of 0.88. These findings not only validate the effectiveness of the proposed approach in the field of corn yield prediction but also highlight the positive role of multisource data fusion in enhancing the performance of prediction models.

With increased heat control requirements for high-heat-flux products in a narrow heat dissipation space, the ultra-thin micro-heat pipe (MHP) with high heat transfer performance has become an ideal heat dissipation component. In this study, the computational fluid dynamics (CFD) method is used to conduct three-dimensional modeling based on the geometric structure characteristics of an ultra-thin MHP. The capillary pressure of the sintered wick is represented by the modified parameter, and a simple and valuable heat and mass transfer model of the ultra-thin MHP is established by fitting the real experimental data through parameter modification. The flow situation of the working medium inside the ultra-thin MHP is analyzed based on the abovementioned parameters. The results show that when the modified parameter is α = 1.5, the temperature equalization requirements of the ultra-thin MHP can be met to the best degree. Moreover, with an increase in heating power, the error value between the surface temperature data of the model and the experimental data of the ultra-thin MHP sample decreases. Under different heating powers, the working medium inside the ultra-thin MHP has the same flow trend. In addition, a 40% increase in temperature difference is found at the junction of the heating section and the adiabatic section, leading to a fluctuation in the temperature gradient on the heat pipe surface. The research results provide a theoretical basis for the model establishment, heat and mass transfer performance investigation, and parameter optimization of ultra-thin MHPs.

This study constructs a multi-stage hybrid forecasting model using hog price time series data and its influencing factors to improve prediction accuracy. First, seven benchmark models including Prophet, ARIMA, and LSTM were applied to raw price series, where results demonstrated that deep learning models significantly outperformed traditional methods. Subsequently, STL decomposition decoupled the series into trend, seasonal, and residual components for component-specific modeling, achieving a 22.6% reduction in average MAE compared to raw data modeling. Further integration of Spearman correlation analysis and PCA dimensionality reduction created multidimensional feature sets, revealing substantial accuracy improvements: The BiLSTM model achieved an 83.6% cumulative MAE reduction from 1.65 (raw data) to 0.27 (STL-PCA), while traditional models like Prophet showed an 82.2% MAE decrease after feature engineering optimization. Finally, the Beluga Whale Optimization (BWO)-tuned STL-PCA-BWO-BiLSTM hybrid model delivered optimal performance on test sets (RMSE = 0.22, MAE = 0.16, MAPE = 0.99%, R2=0.98 ), exhibiting 40.7% higher accuracy than unoptimized BiLSTM (MAE = 0.27). The research demonstrates that the synergy of temporal decomposition, feature dimensionality reduction, and intelligent optimization reduces hog price prediction errors by over 80%, with STL-PCA feature engineering contributing 67.4% of the improvement. This work establishes an innovative “decomposition-reconstruction-optimization” framework for agricultural economic time series forecasting.

Crop diseases threaten agricultural productivity, creating a need for recognition models that combine accuracy, robustness, and computational efficiency. Existing CNN-, Transformer-, and state-space-based approaches can weaken fine-grained lesion edges and textures when emphasizing global context. We therefore propose MambaSeed, a lightweight hybrid network built on MambaVisionMixer and equipped with the Local Continuous Branch Injection mechanism, hereafter LCBI. This mechanism repeatedly introduces local feature compensation during hierarchical modeling, preserving high-frequency cues while maintaining global dependency modeling. On the Tomato, Corn, and Potato datasets, MambaSeed achieved accuracies of 98.97%, 98.06%, and 95.87%, respectively, with 1.13M parameters and 0.240G MACs. In the batch-one FP32 GPU benchmark, its latency was 3.39 ms/image on an NVIDIA RTX 4090. Without a hardware accelerator, the median CPU latency was 25.35 ms/image with one thread and 15.61 ms/image with four threads. Ablation, cross-architecture frequency analysis, and controlled perturbation tests further showed that LCBI strengthens high-frequency feature retention and maintains stable recognition under visual degradation. Together, these results show that MambaSeed provides a compact balance of accuracy, efficiency, and robustness across both accelerator-based and CPU-only inference settings.