To address challenges such as the complex correlations among multiple parameters during the modeling process of plant factory transplanters, the large differences between simulations and actual models, and the difficulties in data acquisition and processing, this paper proposes the concept of a Plant Factory Transplanter (PFT) digital twin five-dimensional model based on research of plant factory transplanters. The PFT digital twin five-dimensional model builds on traditional 3D modeling and includes physical entities, virtual models, services, twin data, and connecting interactions. This study delves deeply into the connotations and construction methods of the PFT five-dimensional model from the five aspects of PFT physical entity, virtual entity, services, twin data, and connections, and illustrates the implementation steps and effects of each link. Finally, practical examples of the application of the PFT digital twin five-dimensional model are presented in actual scenarios. The five-dimensional modeling approach for plant factory transplanters based on digital twins can monitor the working status of transplanters online and evaluate the effectiveness of transplantation. This method overcomes problems such as poor adaptability and difficulty in updating physical models, thus improving the efficiency of monitoring and optimizing configuration parameters. Moreover, the generated virtual entities are more intuitively reflected in the control interface, significantly reducing the reliance of equipment operators on relevant professional skills. In the future, the proposed digital twin five-dimensional model is expected to be further refined and optimized, with creation tools and application scenarios studied. Application research will also be conducted to meet different application requirements.

The plant factory transplanter is a key component of the plant factory system. Its operation status directly affects the quality and survival rate of planted seedlings, which in turn affects the overall yield and economic efficiency. To monitor the operation status and transplanting quality of a transplanting machine in a timely manner, the primary task is to use a computerized and easy-to-use method to monitor the transplanting units. Inspired by the latest developments in augmented reality and robotics, a digital twin model-based and data-driven online monitoring method for plant factory transplanting equipment is proposed. First, a data-driven and virtual model approach is combined to construct a multi-domain digital twin of the transplanting equipment. Then, taking the vibration frequency domain signal above the transplanting manipulator and the image features of the transplanting seedling tray as input variables, the evaluation method and configuration method of the plant factory transplanter digital twin system are proposed. Finally, the effect of the transplanter is evaluated, and the cycle can be repeated to optimize the transplanter to achieve optimal operation parameters. The results show that the digital twin model can effectively use the sensor data to identify the mechanical vibration characteristics and avoid affecting transplanting quality due to mechanical resonance. At a transplanting rate of 3000 plants/h, the transplanting efficiency can be maintained at a high level and the vibration signal of the X, Y, and Z-axis above the transplanting manipulator is relatively calm. In this case, Combined the optimal threshold method with the traditional Wiener algorithm, the identification rate of healthy potted seedlings can reach 94.3%. Through comprehensively using the optimal threshold method and 3D block matching filtering algorithm for image threshold segmentation and denoising, the recognition rate of healthy seedlings has reached over 96.10%. In addition, the developed digital twin can predict the operational efficiency and optimal timing of the detected transplanter, even if the environmental and sensor data are not included in the training. The proposed digital twin model can be used for damage detection and operational effectiveness assessment of other plant factory equipment structures.

In response to the problem of imprecise maintenance of plant spacing and row spacing during corn sowing operations in the middle of existing corn farms, this study designed a dynamic compensation system for corn seed dropping position. The test system includes a corn seeding machine operating unit, a seed dropping variable measurement and control system unit, a terminal control system, and a dynamic compensation system. The dynamic performance of the control system was tested experimentally. When undisturbed signals were loaded, regardless of whether the machine used the position compensation function or not, the coefficient of variation of longitudinal grain spacing was 2.6% ± 0.2%, and the longitudinal grain spacing was basically consistent. After adding disturbance signals, compared with not using the dynamic compensation function, the variation coefficient of seed spacing decreased by an average of 6.94% when using the dynamic compensation function. When using the dynamic compensation function, compared with not using dynamic compensation function, the variation coefficient of lateral grain spacing decreased by an average of 9.16%. Stability analysis of longitudinal and lateral grain spacing was conducted through bench tests, verifying the stability of the proposed dynamic compensation system for seed dropping position.

This work aims to construct an artificial neural network (ANN) ant colony algorithm (ACA)-based fine recognition system for plant factory seedling phenotypes. To address the problems of complexity and high delay of the plant recognition system in plant factories, first, multiple cameras at different positions are employed to collect images of seedlings and construct 3D images. Then, the mask region convolutional neural networks (MRCNN) algorithm is adopted to analyze plant phenotypes. Finally, the optimized ACA is employed to optimize the process timing in the plant factory, thereby constructing a plant factory seedling phenotype fine identification system via ANN combined with ACA. Moreover, the model performance is analyzed. The results show that plants have four stages of phenotypes, namely, the germination stage, seedling stage, rosette stage, and heading stage. The accuracy of the germination stage reaches 97.01%, and the required test time is 5.64 s. Additionally, the optimization accuracy of the process timing sequence of the proposed model algorithm is maintained at 90.26%, and the delay and energy consumption are stabilized at 20.17 ms and 17.71, respectively, when the data volume is 6000 Mb. However, the problem of image acquisition occlusion in the process of 3D image construction still needs further study. Therefore, the constructed ANN-ACA-based fine recognition system for plant seedling phenotypes can optimize the process timing in a more real-time and lower energy consumption way and provide a reference for the integrated progression of unmanned intelligent recognition systems and complete sets of equipment for plant plants in the later stage.

In response to the problem of bowl seedling detachment caused by the shaking of the transplanting machine in plant factories, this paper proposes a physical entity monitoring method for the digital twin (DT) plant factory transplanting system. The method is used to analyze the vibration signals of the transplanting machine under different operating conditions and explore the optimal working conditions. Firstly, a demand analysis for the physical entity of the DT plant factory transplanting system is conducted, focusing on practical applications. Then, an optimal deployment plan is designed based on the axiomatic design (AD) theory. Subsequently, a comparative analysis of the operating conditions of the plant factory transplanting equipment is carried out using data-driven approaches. Finally, the optimal working condition parameters are determined by comparing the modal vibration power spectral density of the transplanting equipment under different operating conditions. The results show that the maximum amplitude occurs in the Z-axis, with a magnitude of 2.1 m/s². By comparing the cloud maps, it is evident that the vibration trends in the Z-axis and X-axis above the transplanting robotic arm are more pronounced compared to the Y-axis. This indicates that under the operating condition of transplanting 3000 plants per hour, a high transplanting efficiency can be maintained, and the vibration signals in the XYZ-axis above the transplanting robotic arm are relatively smooth, making them suitable for transplanting operations. This study combines digital twin technology to analyze the vibration signals of the plant factory transplanting machine under different operating conditions and explore the optimal working conditions. Compared to traditional monitoring platforms, this method facilitates the real-time visualization of different operating conditions of the transplanting machine in a virtual mapping, providing a more intuitive reflection of the equipment operation status.

Modal parameter analysis is a common method for structural characteristic analysis. To accurately identify the modal parameters that describe the vibration characteristics of the structure, the chassis frame of the rice transplanter is taken as the research object. The least squares complex frequency domain method, the admittance circle method, and the modal peak picking method are used to process the measured vibration signal. The modal parameters, such as the natural frequency, damping ratio, and modal shape of the chassis frame, are obtained. Except for the eight-order frequency error greater than 10%, the errors of all other orders are all less than 10%. In the modal test of the frame structure of the rice transplanter, the identification results of the modal parameters are generally reliable. The recognition accuracy of the PolyLeast-Squares Complex Frequency (PolyLSCF) algorithm is lower than the recognition accuracy of the admittance circle method and modal peak picking. The values of some off-diagonal elements in the modal mass matrix calculated by the PolyLSCF algorithm and the admittance circle method are greater than 0.2. The diagonal elements of the modal mass matrix calculated by the modal peak picking method are all 1.