Green apples exhibit a coloration that closely matches their surrounding environment, leading to low recognition accuracy for existing artificial intelligence models. This paper presents a green apple recognition algorithm that integrates an improved U-shaped network (U-Net) and you only look once network (YOLO) v8 to address this challenge. First, the U-Net is enhanced via Dilated Convolution, Attention Gates, and Residual Connections to blur the background, thereby emphasizing the green apple target. Second, convolutional transformations and an attention mechanism are incorporated into YOLO v8, enabling it to focus more effectively on green apple targets within similarly colored backgrounds. Finally, the improved YOLO v8 is employed to recognize green apple targets segmented by the U-Net, with its performance compared against existing models. Research results show that the proposed algorithm achieves a precision of 92.5% and a Recall of 96.8% in green apple recognition, representing a significant improvement over classical models. To mitigate omission issues and further enhance overall performance, an improved YOLO v8 module is connected in parallel with the primary model. Based on its underlying principles, this approach is also applicable to other green fruits with colors and textures highly similar to their backgrounds, demonstrating strong robustness and generalization capabilities.

The hydro-mechanical transmission (HMT) of continuously variable transmission tractors typically achieves speed regulation using a 2Z-X(A) planetary gear system. Long-term use of this setup has created a strong patent barrier, hindering further HMT structural innovation. This study systematically examines the energy consumption characteristics of HMTs based on a 2Z-X(B) planetary gear set configuration, aiming to provide a theoretical reference for developing new HMT tractors. First, the powertrains of both 4-range and 2-range HMTs using this configuration are described. Next, a mathematical model of the 4-range HMT is developed, and its hydrostatic power portion, transmission efficiency, and fuel consumption are analyzed. Finally, the energy consumption characteristics of the 2-range HMT are compared with those of the 4-range HMT, highlighting their performance differences. Results indicate that HMTs based on the 2Z-X(B) planetary gear set exhibit similar efficiency characteristics to traditional systems, with a maximum efficiency exceeding 90%. The impact of efficiency on HMT fuel economy is greater than that of engine fuel consumption itself, suggesting that an efficiency-prioritized power matching control strategy is feasible.

The traction performance analysis of hydro-mechanical transmission (HMT) tractors is employed to evaluate and optimize the transmission system during the design phase, thereby reducing research and development costs associated with continuously variable tractor transmission systems. However, there is currently no established methodological framework for calculating the traction performance of HMT tractors. To address this gap, this study integrated traditional tractor traction performance calculation equations with a self-developed HMT energy consumption calculation method, thereby developing a traction performance calculation model for HMT tractors. Initially, the principle of the tractor’s hydrostatic power-split transmission system was introduced. Subsequently, mathematical models for calculating transmission system energy consumption and tractor traction performance were established, with key sub-models experimentally validated to ensure the reliability of subsequent results. On this foundation, a calculation method for the traction performance of HMT tractors was proposed. Finally, models before and after energy consumption optimization, as well as models under different road conditions and transmission system configurations, were utilized as comparative models to calculate the traction performance of HMT tractors under various settings. The results indicate that energy optimization of HMT not only improves the transmission performance and reduces fuel consumption of HMT tractors, but also enhances the matching capability of HMT tractors with farm tools under high traction efficiency. Additionally, road conditions significantly impact the traction performance of HMT tractors. In wheat stubble fields, the tractor’s traction performance is substantially lower than on standard roads, with a maximum traction force decrease of 40.8% at a slip rate of 30%.

Precise and real-time posture recognition of scallions during harvesting and post-harvest processing is critical for automated conveying, orientation adjustment, bundling, and packaging. Nevertheless, their slender and flexible form, varied spatial orientations, target overlap, lighting fluctuations, background interference, and constrained computational resources of edge devices present significant obstacles to reliable visual perception. This research introduces GOPD-YOLO, a lightweight oriented object-detection network built on the YOLOv8-OBB framework. The network integrates partial-convolution-based lightweight feature extraction to minimize redundant computation, large separable-kernel attention to boost long-range structural representation, and a shared detail-enhanced detection head to improve boundary- and orientation-sensitive prediction. A custom dataset comprising 1500 conveyor-belt images under diverse scallion posture scenarios was developed for model training and assessment. GOPD-YOLO attained a precision of 90.6%, a recall of 94.5%, an mAP@0.5 of 93.2%, and an mAP@0.5:0.95 of 71.5%, with 2.38 million parameters, 6.6 GFLOPs, and a model size of 4.9 MB. Relative to YOLOv8n-OBB, GOPD-YOLO enhanced recall by 3.8 percentage points while decreasing parameter count and model size by 22.7% and 22.2%, respectively. Deployment tests were performed on the Jetson Orin NX Super platform across varying conveyor speeds, lighting conditions, and scallion stacking levels to evaluate the model’s practical utility. These results indicate GOPD-YOLO’s potential as a lightweight vision-based solution for scallion posture recognition in automated harvesting and post-harvest processing.

The power-shift system employed in agricultural multi-range continuously variable transmissions (CVTs) features a complex structure and control logic, presenting significant challenges to the reliability of agricultural machinery. To enable timely detection of faults, constructing an intelligent fault diagnosis classifier to monitor the system’s health status is essential. Typically, fault samples utilized for classifier development originate from ideal bench tests, characterized by uniform patterns and limited diversity, thereby hindering the algorithm’s generalization capability. This study addresses this issue by proposing a generative adversarial network (GAN) model, integrated with a triple loss function and a novel generator architecture, to augment the fault dataset under laboratory conditions. The generator architecture comprises a variational autoencoder module and an oil pressure point attention mechanism, enabling the generation of diverse and fluctuating virtual samples. Building on this augmented dataset, a fault classifier based on one-dimensional ConvNeXt was developed. Experimental results indicate that the classifier achieves an accuracy of 99.73%. While classifier accuracy decreases with increasing noise levels, the GAN-generated dataset provides more comprehensive training, resulting in an accuracy approximately 3% higher than that achieved using the original dataset.

Accurate bee counting is important for colony monitoring, pollination assessment, and precision beekeeping, but manual counting and dense point annotation are labor-intensive. This study proposes M3DANet, a lightweight semi-supervised density regression network with a handheld edge deployment system for bee colony counting. A dataset containing 586 valid high-resolution images and 34,869 point annotations was constructed for training and evaluation. M3DANet uses the first seven stages of MobileNetV3-Large as the lightweight backbone and combines multi-scale context encoding, attention-guided low-level feature fusion, and teacher–student consistency learning with confidence masking and warm-up training. The 10%, 30%, and 50% labeled data settings refer to the proportions of labeled images in the training set, and the remaining training images are used as unlabeled data. Mean absolute error (MAE) and root mean square error (RMSE) are used as evaluation metrics. On the main dataset, M3DANet achieved MAE values of 9.937, 7.003, and 5.570 and RMSE values of 13.093, 9.387, and 7.620 under the 10%, 30%, and 50% settings, respectively, outperforming representative semi-supervised baselines. Under the fully supervised setting, it achieved an MAE of 5.201 and an RMSE of 6.989 with only 2.095 M parameters and 416.64 FPS, using 87.1% fewer parameters and running 17.7 times faster than CSRNet. Cross-species experiments confirmed its low-label generalization ability. Jetson Orin NX deployment achieved 65.75 ms/image inference latency and 10.44 FPS complete-pipeline throughput. These results show that M3DANet balances counting accuracy, annotation efficiency, generalization, and edge deployment practicality.