Yueqi Ma, Guohui Fu, Chao Ban, Tong Su, Ruijuan Chi
{"title":"Improved steering model and integrated robust control for a curved path-tracking controller for headland turns","authors":"Yueqi Ma, Guohui Fu, Chao Ban, Tong Su, Ruijuan Chi","doi":"10.1016/j.biosystemseng.2025.104294","DOIUrl":"10.1016/j.biosystemseng.2025.104294","url":null,"abstract":"<div><div>The accuracy of the rice transplanter's curved path-tracking during headland turns significantly impacts the row spacing precision of rice transplanting, especially in the initial stage of each row operation. This study aims to improve the curved path-tracking accuracy of rice transplanters. To achieve this, an improved transplanter steering model and tracking error model are developed based on the transplanter's steering characteristics, using the transplanting arm array centre as the reference point. Building upon the proposed tracking error model, an integrated robust curved path-tracking controller is proposed, combining low-frequency disturbance observer-based feedforward control, Linear Quadratic Regulator control, H-infinity control, and quadratic stability. This controller is robust to perturbation parameters and disturbances caused by uneven paddy field bottom, sideslip, path curvature, and model linearisation, and it also has a rapid convergence rate. Model comparison results indicated that the turning radii predicted by the proposed transplanter steering model closely match the actual turning radii of the rice transplanter, outperforming the conventional Ackermann steering model. Additionally, the controller using the transplanting arm array centre as the reference point exhibited higher tracking accuracy for transplanting arm array centre compared to using the rear axle centre as the reference point. Ablation experiments demonstrated the effectiveness of each component in the proposed control method, among all components, the low-frequency disturbance observer-based feedforward control had the most significant impact on accuracy improvement. Overall, the proposed curved path-tracking controller exhibited high accuracy for curved path-tracking control and effectively meets the operational requirements of the rice transplanter.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"259 ","pages":"Article 104294"},"PeriodicalIF":5.3,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145119302","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Wenqi Zhou, Huaiyu Liu, Yao Wang, Cunliang Liu, Han Tang, Qi Wang, Jinwu Wang
{"title":"Research on intelligent maize targeted fertilisation method based on BPNN PID adaptive position feedback regulation","authors":"Wenqi Zhou, Huaiyu Liu, Yao Wang, Cunliang Liu, Han Tang, Qi Wang, Jinwu Wang","doi":"10.1016/j.biosystemseng.2025.104292","DOIUrl":"10.1016/j.biosystemseng.2025.104292","url":null,"abstract":"<div><div>Given problems, such as low accuracy of fertiliser application control, large positioning errors, and poor fault monitoring effects in targeted fertilisation operations, this study proposes an intelligent maize-targeted fertilisation method based on a Backpropagation Neural Network (BPNN) Proportional-Integral-Derivative (PID) adaptive position feedback regulation. With the STM32 microcontroller as the master-slave controller, an intelligent maize-targeted fertilisation system was developed through the design of multi-sensor fusion, control parameter calculation and optimisation, construction of a fertilisation drive device, and fault monitoring system. BPNN PID adaptive optimisation was used to control the angular displacement of the fertiliser applicator, and automatic control technology drove the targeted fertilisation mechanism. By integrating dual photoelectric sensors to detect the target maize, an encoder collects the angular displacement of the fertiliser applicator, a ranging sensor monitors the fertiliser amount in the fertiliser box, a pressure sensor monitors the status of the fertiliser pipe, a positioning sensor monitors the operation speed, and multi-machine communication processes the fertilisation operation data. Targeted control and fault monitoring of fertilisation operations under multi-sensor fusion were realised. The adjustment time of the optimisation algorithm is 0.9 s, and the response is fast. Experiments show that the accuracy of fertiliser application control is greater than 95 %, the average positioning error of fertilisation is less than 28.1 mm, the fault alarm success rate reaches 97 %, and the average response time of fault alarm is less than 0.45 s. The intelligent maize-targeted fertilisation method in this study can achieve precise fertilisation control in maize-targeted fertilisation operations.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"259 ","pages":"Article 104292"},"PeriodicalIF":5.3,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145119305","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Peng Hou , Tuo Yin , Shengqi Jian , Yan Li , Xinhao Gao , Xueli Zhang , Changjian Ma
{"title":"Improving sediment discharge efficiency in drip emitters via Tesla-inspired microchannels: PyFluent simulation and SHAP-based structural insights","authors":"Peng Hou , Tuo Yin , Shengqi Jian , Yan Li , Xinhao Gao , Xueli Zhang , Changjian Ma","doi":"10.1016/j.biosystemseng.2025.104295","DOIUrl":"10.1016/j.biosystemseng.2025.104295","url":null,"abstract":"<div><div>Sediment deposition is a critical factor contributing to emitter clogging and flow instability in drip irrigation systems, particularly under sediment-laden water conditions. At the micro-scale (10–1000 μm), flow and particle transport within emitter channels are governed by complex interactions involving confinement effects, turbulent structures, and particle–wall interactions. However, the mechanisms controlling sediment migration and removal remain insufficiently understood, and there is a lack of robust modelling tools to support emitter design under such conditions. In this study, a novel Tesla-inspired bidirectional microchannel was proposed to improve hydraulic performance and sediment discharge efficiency. A high-resolution Euler–Lagrange two-phase flow model was developed using PyFluent, integrating key physical processes including Schiller–Naumann drag, Saffman lift, turbulent dispersion, and rebound boundary conditions to simulate sediment behaviour at particle scale. Simulation results revealed that the inclusion of reverse-flow units significantly enhanced shear zones and vortex intensity, leading to a 97.18 % increase in turbulent kinetic energy (TKE, CFD simulation). Under different forward- and reverse-flow unit configurations, PSD and QSDV both decreased by 22.73 %–53.40 %. Variations under different channel widths and depths showed different ranges due to QSDV being normalised by volume (all CFD simulation results). Contribution analysis using SHapley Additive exPlanations (SHAP) identified hydraulic diameter and the number of forward-flow units as dominant structural factors influencing sediment transport through their effects on local energy dissipation and flow field reorganisation. These findings provide a physically interpretable and practically applicable modelling framework for optimising emitter design. This study proposed approach offers new insights into the coupling between microchannel geometry and sediment dynamics, supporting the development of anti-clogging strategies in drip irrigation systems using non-conventional water sources.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"259 ","pages":"Article 104295"},"PeriodicalIF":5.3,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145119306","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Li Du , Yang Xu , Can Wang , Yiyi Chen , Junhua Zhang , Bin Ma , Danyan Chen
{"title":"An improved light efficiency LED array design via increasing uniformity for pea sprouts","authors":"Li Du , Yang Xu , Can Wang , Yiyi Chen , Junhua Zhang , Bin Ma , Danyan Chen","doi":"10.1016/j.biosystemseng.2025.104290","DOIUrl":"10.1016/j.biosystemseng.2025.104290","url":null,"abstract":"<div><div>Light distribution is an important factor affecting plant growth under facility lighting. However, traditional lighting provides uneven illumination, which increases the difficulty of cultivation management and creates challenges in standardised plant production. Achieving the best light environment is highly significant for high-efficiency production. To effectively address uneven light distribution in the planting layer, this study proposed a high light efficiency LED array design that increases light distribution uniformity based on the improved genetic algorithm. This design has been verified by optical simulation, spectrometer measurement and pea sprout cultivation. The simulation results showed that the illumination uniformity of the optimised LED array and traditional square LED array were 91.7 % and 85.7 %, respectively. The uniformity of the illuminance measured by the spectrometer for two LED arrays was 92.5 % and 80.2 %, respectively. Furthermore, the effects of the two LED arrays on the growth of pea sprouts were compared. The total light intensity of the optimised LED array was reduced by 20.1 % lower; however, the yield of the pea sprouts increased by 8.6 % higher. The light intensity required to produce pea sprouts per unit mass was 26.5 % lower, while the energy and economic efficiency improved. Therefore, the LED array designed based on the improved genetic algorithm has high illumination uniformity and light use efficiency and presents a novel method for improving pea sprout production and lighting optimisation strategy.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"259 ","pages":"Article 104290"},"PeriodicalIF":5.3,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145119303","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Tarl M. Berry , Nurayn A. Tiamiyu , Jacques van Zyl , Umezuruike L. Opara , Paul Cronje , Alemayehu Ambaw , Vaughan Hattingh , Corné Coetzee , Thijs Defraeye
{"title":"Fruit cooling performance analysis within a fully loaded refrigerated container: CFD modelling and validation","authors":"Tarl M. Berry , Nurayn A. Tiamiyu , Jacques van Zyl , Umezuruike L. Opara , Paul Cronje , Alemayehu Ambaw , Vaughan Hattingh , Corné Coetzee , Thijs Defraeye","doi":"10.1016/j.biosystemseng.2025.104254","DOIUrl":"10.1016/j.biosystemseng.2025.104254","url":null,"abstract":"<div><div>Refrigerated containers (RCs) are crucial for transporting fresh produce to international markets, significantly influencing fruit quality along the cold chain. Although RCs are a mature technology, fresh produce industries report challenges relating to temperature heterogeneity and ineffective monitoring approaches. This study developed and validated a computational fluid dynamics model to characterise airflow and heat transfer inside an RC packed with standard ventilated packaging for South African citrus fruit. The model was implemented with an accurate representation of a refrigeration unit, incorporating the presence of fans and evaporator coils based on experimental characterisations. Air speed and cooling validations showed good agreement with the models. Simulations identified a vertically dominant airflow pattern, with air speeds within the pallets ranging from 0.03 m s<sup>-1</sup> to 0.16 m s<sup>-1</sup>. Air velocities within the pallet regions were categorised into four zones: a turbulent air recirculation zone, a high-velocity stabilisation zone, a declining air velocity zone, and a heterogeneous air velocity zone near the door. Excessively cooled regions were identified, potentially increasing chilling injury risk, a primary concern for South African citrus exports. The study evaluated temperature monitoring, and an optimal position for hygrothermal sensors was proposed for single-device monitoring. It was further shown that air temperature data is conditionally representative of pulp temperature. The insights gained can guide industry practitioners in enhancing temperature monitoring practices and inform future research on optimising RC cooling efficiency and minimising chilling injury risks to improve fruit quality and reduce waste in the fresh produce supply chain.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"259 ","pages":"Article 104254"},"PeriodicalIF":5.3,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145099584","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yipeng Cui , Pengxuan Guan , Jianning Yin , Zehao Zha , Qiming Yu , Zhenwei Wang , Duanyang Geng
{"title":"Influence of maize picking roller surface structure on stalk pulling force","authors":"Yipeng Cui , Pengxuan Guan , Jianning Yin , Zehao Zha , Qiming Yu , Zhenwei Wang , Duanyang Geng","doi":"10.1016/j.biosystemseng.2025.104289","DOIUrl":"10.1016/j.biosystemseng.2025.104289","url":null,"abstract":"<div><div>During the maize ear harvesting process, a reasonable selection of the picking roller's surface structure can significantly enhance stalk pulling force, reduce ear-picking losses, and improve overall harvesting efficiency. Investigating the influence of different picking roller surface structures on stalk pulling force is therefore of critical importance. In this study, a simulation model was developed based on the Discrete Element Method (DEM) and Multi-Body Dynamics (MBD) to simulate the interaction mechanisms between the ear-picking device and maize stalks. The accuracy of the simulation model was validated through bench tests, using maximum stalk pulling force and power consumption as key evaluation metrics, with relative errors of 5.4 % and 5.2 %, respectively. The study further explored the effects of picking roller surface structure (pattern shape, pattern height and pattern spacing) on stalk pulling force. The results indicate that pattern shape, pattern height, pattern spacing, and their interactions have a significant impact on stalk pulling force. The optimal surface structure of the picking roller is a inclined pattern structure with a pattern height of 2.5 mm and a pattern spacing of 8 mm. The simulation results can be used to analyse the effect of the picking roller surface structure on stalk pulling force, providing a theoretical basis for the rational selection of picking roller surface structures.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"259 ","pages":"Article 104289"},"PeriodicalIF":5.3,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145099596","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Angela Ramon-Perez , Irene Camerlink , Nienke van Staaveren , Kristina Maschat , Kenny van Langeveld , Thomas Banhazi , Michaela Fels , Maite Jachens , Jarissa Maselyne , Björn Forkman , Pol Llonch
{"title":"Technologies for automatic assessment of pig welfare using animal-based indicators in the slaughterhouse: a review","authors":"Angela Ramon-Perez , Irene Camerlink , Nienke van Staaveren , Kristina Maschat , Kenny van Langeveld , Thomas Banhazi , Michaela Fels , Maite Jachens , Jarissa Maselyne , Björn Forkman , Pol Llonch","doi":"10.1016/j.biosystemseng.2025.104286","DOIUrl":"10.1016/j.biosystemseng.2025.104286","url":null,"abstract":"<div><div>Most meat-producing species end their life at the slaughterhouse. Here, animals are gathered from diverse farms, allowing for extensive data collection, including on welfare status. Assessing animal welfare requires reliable indicators, particularly those that are animal-based. Automated welfare evaluation offers a continuous, objective, and consistent approach for monitoring large numbers of animals, eliminating human bias and fatigue associated with high-speed production lines, and decreasing farm visits. This review aims to identify animal-based welfare indicators for pigs that can be automatically measured at slaughterhouses and to examine commercially available Precision Livestock Farming (PLF) technologies used at the slaughterhouse, including prototypes and on-farm technologies that can be adapted and applied to slaughterhouses. A three-step methodology is used: first a systematic literature search, followed by a comprehensible commercial search, and finally an expert consultation survey to confirm that all technologies were identified. A total of 16 technologies for slaughterhouse applications and 71 technologies for on-farm use were identified. Among the on-farm technologies, 52 were deemed feasible for slaughterhouse implementation, while 19 were considered unsuitable due to mismatches with slaughterhouse purposes, such as feeding behaviour or heat detection. The results also highlight the need to address automated welfare assessment during the transport phase to ensure thorough understanding and continuous monitoring of animal welfare across the entire production chain. While automated systems for monitoring pig welfare show significant potential, challenges in practical implementation and widespread adoption remain, requiring collaboration between researchers, industry stakeholders, and technology developers to fully realise their potential.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"259 ","pages":"Article 104286"},"PeriodicalIF":5.3,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145099598","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Vinay Vijayakumar , Yiannis Ampatzidis , Christian Lacerda , Tom Burks , Won Suk Lee , John Schueller
{"title":"AI-driven real-time weed detection and robotic smart spraying for optimised performance and operational speed in vegetable production","authors":"Vinay Vijayakumar , Yiannis Ampatzidis , Christian Lacerda , Tom Burks , Won Suk Lee , John Schueller","doi":"10.1016/j.biosystemseng.2025.104288","DOIUrl":"10.1016/j.biosystemseng.2025.104288","url":null,"abstract":"<div><div>For effective weed control in vegetable farms, enhancing precision spraying through improved real-time detection is crucial. Over the years, weed detection studies have evolved from traditional feature-based methods to deep learning approaches, particularly convolutional neural networks (CNNs). While numerous studies have focused on improving detection accuracy by experimenting with different backbones, architectures, and hyperparameter tuning, fewer have addressed the real-time implementation of these models in field conditions. Existing research primarily benchmarks model inference speed but often neglects the broader algorithmic efficiency, which includes sensor data integration, processing pipelines, and microcontroller output handling. Furthermore, real-world deployment challenges, such as camera performance at different robot speeds, the optimal operational range for high detection accuracy, and the end-to-end latency of the machine vision system, remain underexplored. This study addresses these gaps by training a custom YOLOv8 nano model to detect three weed types (broadleaf, nutsedge, and grass) and two crop types (pepper and tomato) in plasticulture beds. The system runs on a robotic smart sprayer in real time, integrating GPS and camera data while transmitting control signals to the microcontroller. Beyond detection performance, we evaluate the entire processing pipeline by measuring the total loop time and its variation with the number of detections per frame. Additionally, the optimal robot operational speed was determined, finding that 0.45–0.89 m s<sup>−1</sup> provides the best balance between detection accuracy and system responsiveness. By focusing on end-to-end real-time performance on vegetable beds, this study provides insights into the practical deployment of smart spraying, often been overlooked in prior research.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"259 ","pages":"Article 104288"},"PeriodicalIF":5.3,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145099583","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Parameter optimisation of a centrifugal fan for rice combine harvesters based on airflow resistance coefficients and CFD simulations","authors":"Zhenwei Liang , Million Eyasu Wada","doi":"10.1016/j.biosystemseng.2025.104287","DOIUrl":"10.1016/j.biosystemseng.2025.104287","url":null,"abstract":"<div><div>In this work, both experimental and numerical simulations were employed to identify optimal fan parameter settings for achieving efficient cleaning performance during high-yield rice harvesting. First, field experiments were conducted to analyse the distribution of threshed outputs within the cleaning shoe, and then airflow resistance coefficients created by the fluidised grain and cleaning sieves in each zone were calculated. Subsequently, perforated plates were designed based on the calculated airflow resistance coefficients in different sieve zones to represent the cleaning load. The computational fluid dynamics (CFD) simulation results were validated by using measured airflow velocity at multiple points beneath the perforated plates. After validation, additional CFD simulations were performed under various fan parameter settings, incorporating porous media to simulate the fan's working load. The results indicated that a sieve opening of 26 mm, guide plate angles (I) of 38° and (II) of 36°, and a fan speed of 1300 rpm significantly improved airflow and pressure distribution within the fan. Finally, a field experiment validated the cleaning performance using the selected parameter combinations, achieving a grain sieve loss ratio of 0.78 % and a grain impurity ratio of 1.15 % at a feed rate of 6 kg s<sup>−1</sup>. This innovative approach not only provides an accurate method for determining the fan's working load but also enables the evaluation of fan performance under varying load conditions through CFD simulations, ultimately enhancing the cleaning performance of rice combine harvesters through optimised parameter selection.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"259 ","pages":"Article 104287"},"PeriodicalIF":5.3,"publicationDate":"2025-09-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145047041","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Application of weight prediction for Holstein dairy cows in non-pregnant and postpartum stages","authors":"Hsin-I Chiang , Jia-Ming Zhou , Wen-Lin Chu","doi":"10.1016/j.biosystemseng.2025.104276","DOIUrl":"10.1016/j.biosystemseng.2025.104276","url":null,"abstract":"<div><div>A non-contact weight prediction system for Holstein dairy cows was developed based on depth sensing technology, designed to predict weight changes during non-pregnant and postpartum stages. The system utilises an Intel RealSense D455 depth camera to capture depth image information from cow's dorsal, hips, and side regions, extracting effective body surface feature data through a systematic data processing workflow. Experimental results demonstrate that the Gaussian Process Regression (GPR) model performed most excellently in the cow's dorsal region. For example, with cow number cid603 during the non-pregnant period, prediction accuracy reached a root mean square error (RMSE) of 19.37 kg and a mean absolute percentage error (MAPE) of 1.82 %; with cow number cid700 in the postpartum stage, the model maintained an RMSE of 22.35 kg and MAPE of 2.74 %, exhibiting robust model generalisation capability. Compared to traditional farm methods based on body length and heart girth measurements, the weight prediction system proposed in this study significantly improved the accuracy and stability of weight prediction, especially in capturing physiological state changes (such as postpartum weight loss). Experimental results indicate that the GPR model exhibited the best predictive ability and generalisation with feature data from the dorsal region, effectively supporting precise monitoring of dairy cow weight. Future research directions should focus on optimising image preprocessing techniques, incorporating more physiological parameters (such as feed intake), and integrating depth information from different angles to enhance the system's adaptability in complex environments, thereby strengthening the universality and reliability of the weight prediction model.</div></div>","PeriodicalId":9173,"journal":{"name":"Biosystems Engineering","volume":"259 ","pages":"Article 104276"},"PeriodicalIF":5.3,"publicationDate":"2025-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145047040","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}