Tomato is one of the most widely grown vegetables in the world. However, many challenges still remain in the actual cultivation in the broad market prospects of tomato production. Among them, the pruning of branches and leaves has been one of the most important steps during tomato growth and fruiting. However, manual pruning cannot fully meet the large-scale production in recent years, due to the high labor intensity and cost. Particularly, interval pruning is often required to work consistently for long periods of time. Fortunately, the tomato pruning robot can be expected to work all the whole day and night. It is an urgent need for the tomato pruning robots to accurately and efficiently identify the tomato stems and branches. Tomato stems and branches can often be recognized well during the daytime. But the low accuracy and missed detection of tomato branches can occur in a night environment at present. In this study, a segmentation model (NTS-YOLO) was proposed for the tomato stems and branches in night environment using improved CycleGAN and YOLOv8. The feature extraction of the CycleGAN module was improved to solve the color distortion and blurring of images. The low- and high-level features were then fused to introduce the efficient channel attention (ECA) mechanism in the network. The YOLOv8 backbone network was replaced by the lightweight backbone network (StarNet), in order to reduce the complexity of the improved model. The YOLOv8 neck network was also replaced by the Gold-YOLO. The convolutional block attention module (CBAM) attention mechanism was embedded in the head network, in order to improve the accuracy of the improved model. The results showed that the FID, LPIPS of the images generated by the improved CycleGAN were reduced by 12.23 and 0.07 and PSNR increased by 2.96 dB, respectively, compared to the original CycleGAN model. The NTS-YOLO improved mAP by 19.8 percentage points using the data-enhanced datasets. The ablation experiments indicated that the mean values of precision, recall, and average accuracy of the NTS-YOLO model were achieved at 92.5%, 86.1%, and 93.3%, respectively, which were improved by 3.8, 2.4, and 4.5 percentage points, respectively, compared with the original network. The frame rate of detection increased from 70.9 to 75.3 frame per second. The effectiveness of the NTS-YOLO model was validated using ablation experiments. The NTS-YOLO model achieved 95.3%, 92.4%, and 92.2% in the AP of the stem, the lateral branch, and the fruit branch, respectively, which were improved by 4.1, 4.1, and 5.3 percentage points, respectively, compared with the original network. Furthermore, the mean average accuracy of the NTS-YOLO model reached 93.3%, which was an increase of 15.0%, 18.8%, 5.7% and 4.5%, respectively, compared with the mainstream segmentation models, such as Mask R-CNN, YOLACT, YOLOv5l-seg and YOLOv8l-seg. The leakage rate reached 4.2%, which was reduced by 15.9, 18.2, 10.5, and 5.9 percentage points, respectively. The FPS reached 75.3 frame per second, which was faster than the rest networks by 58.2, 49.4, 2.7 and 4.4 frame per second, respectively. The NTS-YOLO network was more robust and faster than the rest of mainstream segmentation in segmenting tomato stems, lateral branches, and fruit branches in a night environment. This finding can also provide technical support for automatic and intelligent pruning in the tomato-growing industry.

This study aims to determine the rheological properties and structural stability of the beach soil for shellfish cultivation. A series of rheological tests were carried out on the beach soil with different water content and standing time using a rotary rheometer. A single-factor test was also conducted to explore the influence of water content and standing time on the rheological properties of the beach soil. The soil shear was then analyzed to determine the influence of water content and standing time on yield stress. At the same time, the numerical model was established for the yield stress, water content and resting time after the central composite design (CCD) test in the response surface method (RSM). A comparison was then made on the influence of water content and resting time on yield stress. Finally, the reliability and accuracy of the regression model were further verified to compare the error between the measured and the actual values. The results show as follows: 1) The shear stress gradually decreased with the increase of water content under the same shearing rate in the shearing process, whereas there was a gradual increase with the increase of standing time; In addition, the beach soil was characterized by the consolidation. The longer the static time was, the longer the solid-liquid transformation was, and the later the solid-liquid node was. The static time shared the greatest impact on the solid-liquid node in the early stage. The longer the static time was, the smaller the impact on the solid-liquid node was. 2) When the average shear stress and yield stress were at the shear rate of 10 -60 s^-1 under the same standing time, the overall shear stress of the tidal flat soil gradually decreased with the increase of water content, thus reducing the structural stability of the tidal flat soil. In addition, the overall shear stress of the tidal flat soil gradually increased under the same moisture content with the increase of static time, indicating the increase in the structural stability of tidal flat soil. The change rate of yield stress with the static time showed that there was a cut-off point of moisture content in the beach soil, which was between 64.5% and 67.0%. In the samples exceeding this moisture content cut-off point, the change rate of yield stress with the static time remained stable after the static time exceeded 31 h, where there was no fluctuation in the yield stress. The yield stress ranged from 2 240 to 4 380 Pa. In addition, the yield stress in the sample of 62.0%-69.5% was between the 0-53 h, and the yield stress between 62.0%-69.5% was between 1 870-5 410 Pa. 3) The P and R² values of numerical models were above 0.05, and 0.953 4, respectively, indicating the high reliability and accuracy. There was a more significant influence of water content on the yield stress, compared with the resting time. However, the interaction between water content and resting time shared no significant influence on the yield stress. The test verified that the measured yield stress was basically consistent with the predicted one, with an error of less than 15.0%. The rheological properties of beach soil can be better characterized as well. The findings can provide a strong reference for the research and optimization of beach shellfish harvesting machinery.

A contact model can be expected to accurately and rapidly characterize the tidal flat soil using the discrete element method (DEM). In this study, the contact model was optimized using the API secondary development function of the discrete element simulation software (EDEM). The target model was selected as the Hertz-Mindlin contact model. Firstly, the plasticity and adhesion were added to the normal contact force. Secondly, the sliding friction force with the relative tangential velocity was input into the tangential contact force. Then, the moist elasto-plastic adhesion (MEPA) model was obtained. The maximum normal pressure and normal adhesion force of the piston pull-out test were selected as the indexes of virtual calibration. Plackett-Burman test showed that there were significant effects of the correction coefficient of sliding friction, Young 's modulus of soil, static friction coefficient between soil particles, rolling friction coefficient between particles, static friction coefficient between particles, and geometry on the maximum normal pressure of the probe. The maximum adhesion force of the probe depended mainly on the adhesion coefficient between particles and the geometry and their interaction. The piston pull-out test showed that the combination of parameters was achieved in the comparison indexes for the virtual calibration of tidal flat soil under target working conditions, including the 0 h settling time, 0 min contact time, and 210 mm/min separation velocity. The virtual calibration of DEM parameters was realized using non-significant and significant parameters. Among them, the non-significant parameters were referred to the existing data, while the significant parameters were continuously adjusted to make the simulated curve of piston pull-out force close to the actual. The MEPA, JKR, EEPA, and Bonding models were compared to characterize the tidal flat soil. The shoveling test platform was then built to verify the accuracy of the models. The tidal flat soil with a thickness of 11 cm was accumulated in the soil box. The digging shovel had a cutting depth of 5cm in the soil box. Two tests were set firstly, where the angle of the digging shovel was adjusted to 25° and 30°, respectively, and the forward speed was 0.08m/s. Then one test was that the digging shovel angle of 20°, and the forward speed of 0.16 m/s. The force was accurately obtained at the sensor of the shoveling device. The EDEM and the multi-dynamics (RecurDyn) software were also selected to simulate the shoveling. The test results showed that the average absolute error of the MEPA model was controlled within 50 N, compared with the actual shoveling resistance. The simulation accuracies of the MEPA model were about 65.957 %, 74.206%, and 59.326 % higher than those of the JKR, EEPA, and the Bonding model, respectively. Compared with the field test, the relative errors of the MEPA model were 5.598% and 6.362% in the soil accumulation thickness and side margins, respectively. Both errors were remained within 10%. Therefore, the MEPA model can be used to better simulate the tidal flat soil. It is of great significance to simulate and optimize the tidal flat device.

Most of the tidal flat shellfish harvesting equipment predominantly employed tracked or wheeled walking devices, which encountered issues such as significant sinking of the driving components during operation, poor load-carrying capacity of the devices, and high maintenance costs in the later stage. Therefore, this paper designed a tidal flat walking device based on the principle of spiral propulsion, which could be used for transportation, and conducted structural design and experimental research on the walking device. The impact of the spiral blade structure on working performance was analyzed by using DEM-MBD(discrete element method-multibody dynamics) coupling simulation technology. A single-factor test was carried out to identify the optimal range for height, thickness, and helix angle of spiral blade on the walking mechanism. Subsequently, a three-factor and three-level orthogonal test method was then used to determine the optimal design parameters of blade. Finally, no-load and load operating performance tests were carried out on an experimental prototype, which was constructed according to similar criteria to evaluate the working performance of the designed running gear. The results indicated the following:1) When all factors were independent, the slip rate of the walking device increased with higher helix angles. Both the traction force and working efficiency of the walking device initially increased and then decreased with higher helix angles. The slip rate of the walking device increased with the thickness increased, while working efficiency and traction coefficient of the walking device followed an initial increase and then decrease pattern. On the other hand, increasing the height-to-diameter ratio resulted in a decrease followed in slip rate of the walking device. The traction coefficient, increased overall with higher height-to-diameter ratios, with working efficiency exhibiting a similar trend. 2) The walking device demonstrated optimal driving performance with a helix angle of 25°, a height of 150 mm, and a thickness of 7.5 mm as determined by orthogonal test. 3) The slip rate of the walking device was 46.92% with no load, 63.48% with a 10 kg load, and 58.35% with a 15 kg load. The subsidence amount was directly proportional to the load, with values of 47.59, 60.09, and 70.22 mm for no load, 10, and 15 kg load respectively, with a proportional coefficient of 1.512. 4) The rated load of the walking device was approximately 8 kg while the ultimate load was around 21.61 kg, which demonstrated its robust carrying capacity in the tidal flat environment. 5) In comparison with the simulation experiments, the max deviation rates of subsidence and slip rate were 24.614% and 4.061% respectively. This affirmed that the simulation results could offer reference for prediction and analysis of walking device performance. These research findings were valuable for the development of more efficient tidal flat harvesting equipment with stronger carrying capacity.