Reliability Modeling of Shoulder Fatigue and Recovery for Warehouse Operators Performing Dynamic Tasks

项目摘要 对电子商务不断增长的需求导致仓库和配送中心的增加， 拥有运营运营所需的劳动力。为了提高效率，公司正在转向部件到部件 用于订单履行的人员系统，以达到每个工人接近500件/小时的工作效率水平。这些系统 创建高度重复性且主要涉及手臂和肩部的人工拣货作业。重复性 在没有充分休息的情况下，长时间进行手臂运动会导致疲劳和不适。 对于肩膀，这可能会导致肌肉骨骼疾病(MSD)。库存搬运工和订单填充员都有 涉及非工作天数的伤害发生率高于平均水平。减少MSD的数量是一个 运输、仓储和公用事业(TWU)理事会和肌肉骨骼健康(MSH)的目标 跨部门的Nora议程。预防MSD取决于有效的工作设计和作息时间表， 最大限度地减少疲劳。然而，目前的做法依赖于为静态肌肉负荷开发的疲劳模型，该模型 没有考虑到拣货员所经历的动态需求。因此，该计划的主要目标是 拟议的项目是能够预测人工挑选订单造成的疲劳和恢复，重点是 部件到人的系统，具有高度重复的肩部工作。第二个目标是将研究成果转化为 通过向从业者提供这些预测模型来实践(R2P)，以使他们能够融入他们的工作 评估和设计实践。这些目标涉及MSH跨部门议程呼吁研究 实时数据与经验证的预测模型的集成，以解决任务和工作中的可变性- 休息周期。这些模型将根据实验室研究期间收集的数据构建。使用中心复合体 设计、疲劳发展将在一系列负荷水平和重复率以及恢复范围内进行评估 疲劳程度将在一系列休息时间范围内进行测量。受试者将完成四个周期的顺序 采摘，由指定的休息期分隔。相关的衡量标准将包括疲劳的主观评级， 来自可穿戴传感器的运动学数据和任务性能。这些措施将统一为疲软 使用函数回归的结果度量。然后，应用可靠性理论对统一结果进行预测 在分别作为退化和反向退化过程的重复疲劳和恢复循环期间， 考虑任务条件、工作人员特征和时间。合作伙伴仓库的现场验证将是 已执行，其中模型预测将与三个订单挑选作业的工人主观评级进行比较。 一旦得到验证，这些模型将被打包成一个基于网络的应用程序，该应用程序将分发给 从业者(输出)，能够预测未来的工人疲劳程度，这比现有的更具信息性 提供工作人员当前状况或风险的快照的方法。应用修订后的模型可以 促进改进的工作场所设计和作业调度，以适应订单挑拣员的能力，这 支持预防肌肉骨骼疾病和改善工人健康的长期目标(成果)。

Project Summary The growing demand for e-commerce has resulted in an increase in warehouses and distribution centers, along with the needed workforce to run the operations. For improved efficiency, companies are shifting to parts-to- person systems for order fulfillment to reach productivity levels near 500 items/hour per worker. These systems create manual order picking jobs that are highly repetitive and primarily involve the arm and shoulder. Repetitive arm movements, performed for prolonged durations without adequate rest, can result in fatigue and discomfort for the shoulder, which can lead to musculoskeletal disorders (MSDs). Both stock movers and order fillers have above average incidence rates of injuries involving days away from work. Reducing the number of MSDs is an objective of the Transportation, Warehousing, and Utilities (TWU) Council and the Musculoskeletal Health (MSH) Cross-Sector NORA Agendas. Preventing MSDs depends on effective job design and work-rest schedules that minimize fatigue. However, current practice relies on fatigue models developed for static muscle loading, which fail to account for the dynamic demands experienced by order pickers. Thus, the primary objective of the proposed project is to enable prediction of fatigue and recovery resulting from manual order picking, focusing on parts-to-person systems with highly repetitious shoulder work. A secondary objective is to translate the research to practice (r2P) by providing practitioners with these predictive models to enable incorporation into their job evaluation and design practices. These objectives address the MSH cross-sector agenda call for research on the integration of real-time data with validated predictive models that address the variability in tasks and work- rest cycles. The models will be constructed from data collected during an in-lab study. Using a central composite design, fatigue development will be evaluated across a range of load levels and repetition rates, and recovery from fatigue will be measured across a range of rest durations. Subjects will complete four periods of order picking, separated by designated rest periods. Dependent measures will include subjective ratings of fatigue, kinematics data from wearable sensors, and task performance. These measures will be unified into a fatigue outcome metric using functional regression. Then, reliability theory will be applied to predict the unified outcome during repeated fatigue and recovery cycles as degradation and inverse degradation processes, respectively, accounting for task conditions, worker characteristics, and time. Field validation at a partner warehouse will be performed, where model predictions will be compared to worker subjective ratings for three order picking jobs. Once validated, the models will be packaged into a web-based application which will be disseminated to practitioners (output), enabling prediction of future worker fatigue levels, which is more informative than existing methods that provide a snapshot of the worker’s current condition or risk. Application of the revised models can facilitate improved workplace design and job scheduling to accommodate the capacities of order pickers, which supports the long-term goals of preventing musculoskeletal disorders and improving worker health (outcome).