The uniformity of the cultivation environment in a micro-plant factory plays a critical role in ensuring the consistent growth of seedlings, and an optimal airflow pattern is the key to maintaining environmental uniformity. This study applied computational fluid dynamics (CFD) modeling to compare the effects of six different ventilation modes on the microclimate within the cultivation space. In cases 1 and 2, the inlet was positioned at the top, while the outlets were located at both the bottom and the top of the side walls. For cases 3 to 6, a side-inlet and side-outlet ventilation system was employed across the three cultivation layers. Case 4 maintained consistent inlet and outlet airflow speeds, whereas cases 3, 5, and 6 featured airflow settings that either increased or decreased progressively from the top layer to the bottom. Notably, case 6 was characterized by a more compact arrangement of cultivation racks within the space, which were positioned closer to the outlet than in the other cases. In case 1, the air inlets were positioned at the top, while the outlets were located on both side walls at the lower layer of cultivation. In contrast, case 6 used a side-inlet and side-outlet ventilation strategy, in which the airflow speed of the inlets decreases progressively from the top to the bottom of the cultivation layers. Additionally, the cultivation racks in case 6 were arranged more compactly and positioned closer to the outlet of the cultivation space. The relative standard deviation (RSD) was used to evaluate the uniformity of the airflow velocity (m/s), temperature (K), and relative humidity (%) within the crop-growing area. The results indicated that, among all the scenarios, case 6 demonstrated the lowest RSD values for the airflow velocity, temperature, and relative humidity, with reductions of 18.34%, 0.12%, and 2.05%, respectively, compared to the control group (case 1). Based on the ventilation design of case 6, a micro-plant factory was developed featuring a bidirectional flow fan, air conditioning, and PWM fans for the coordinated control of air circulation within the seedling cultivation space, along with adjustable cultivation layer heights and shelf spacing. The accuracy of the CFD model for the micro-plant factory was validated with normalized root mean square error (NMSE) for cultivation layer heights of 250 mm, 300 mm, and 350 mm. The NMSE values comparing the simulated and measured results for the airflow velocity, temperature, and relative humidity were found to be 0.032, 0.0020, and 0.0022; 0.031, 0.0021, and 0.0018; and 0.046, 0.0021, and 0.0021, respectively. These findings indicate that the established CFD model can reliably predict the microenvironment within the micro-plant factory.

Existing force-measuring devices lack versatility in studying the dynamic coupling process between the seedling-picking device and the plug seedling pot during automatic transplanting. This research developed a universal force-measuring device featuring a centrally symmetrical clamping needle layout and a simultaneous insertion and clamping mechanism. The force-measuring device enables the flexible adjustment of the number of clamping needles (2/3/4 needles) via a modular structure. It can also modify the insertion depth and angle of the clamping needles to accommodate three specifications of plug seedlings, namely 50-hole, 72-hole, and 128-hole plug seedlings. A real-time monitoring system with dual pull-pressure sensors is integrated to precisely acquire the dynamic response curves of the clamping force (F_J) and the disengaging force (F_N) of the plug seedling pot during the seedling-picking process. Taking water spinach plug seedlings as the research object and combining with EDEM-RecurDyn coupling simulation, the interaction mechanism between the clamping needle and the plug seedling pot was elucidated. The performance of the force-measuring device was verified through systematic force-measuring experiments. The main research findings are as follows: The force-measuring device designed in this study can successfully obtain the mechanical characteristic curve of the relevant seedling plug pot throughout the automatic seedling-picking process. The simulation results show high consistency with the experimental results, indicating that the force-measuring device can effectively reveal the dynamic coupling process between the seedling-picking device and the plug seedling pot. The verification experiment demonstrates that the force-measuring device can effectively quantify the mechanical properties of the of plug seedling pots under different plug seedlings specifications and different clamping needles configurations. Reducing the hole size and increasing the number of clamping needles can effectively decrease the peak value of the disengaging force (F_Nmax). The peak clamping force (F_Jmax) is approximately inversely proportional to the needle number, with the four-needle layout providing the most uniform force distribution. The force-measuring device developed in this study is functional, applicable, and versatile, offering a general force-measuring tool and a theoretical foundation for optimal seedling-picking device design.

The absence of industrially viable, fully automated transplanting systems for grafted melon-watermelon tray plug transplants stems primarily from intrinsic biomechanical constraints—namely hypocotyl brittleness, suboptimal stem erectness, and excessive plug volume—that collectively hinder reliable, non-traumatic automated manipulation. To overcome this bottleneck in mechanical automated transplanting, we propose a theory-informed and empirically validated framework for designing, optimizing, and implementing a pneumatically actuated, row-wise pick-and-place system. Its two-finger, four-needle end-effector employs sequential substrate penetration followed by controlled radial constraint. First, geometric metrology and quasi-static testing determined a mean tray detachment force of 2.99 ± 0.51 N (n = 30); a nonlinear viscoelastic model of peat-vermiculite compression enabled physics-guided parametrization of kinematic and force-control profiles. Second, EDEM-RecurDyn co-simulation integrated with a Box–Behnken design (three velocities × three levels) supported global multi-objective optimization. Performance was quantified via microstructural integrity (bond breakage density) and macrostructural coherence (substrate mass retention). Damage evolution was mapped across the oblique insertion-parallel clamping-vertical extraction sequence. Pareto-optimal settings were: insertion = 400 mm/s, holding = 100 mm/s, lifting = 300 mm/s. Third, a prototype using medical-grade stainless-steel needles (Φ = 3 mm) and a 50 mm-stroke pneumatic cylinder achieved a 96.33 ± 1.53% success rate, 2.33 ± 1.53% seedling injury rate, and 0.67 ± 0.58% plug pot failure rate (n = 300) under standardized conditions. This work provides a theoretical basis for the design and optimization of automated, low-loss transplanting equipment for grafted melon-watermelon tray pot seedlings.