Development of Risk behavior and posture evaluation model for Power electric industry using Motion sensor

Abstract

Safety is a significant concern in the power and energy sector, with accidents like electric shocks, falls, and crushing occurring frequently and resulting in fatalities. As a response, the government enacted the Serious Accidents Punishment Act of 2022, prompting industrial sites to explore various methods to enhance worker safety. One particular field of interest is ergonomics, which focuses on assessing work postures related to musculoskeletal disorders. Methods such as RULA and REBA are used for this purpose. RULA, or the Rapid Upper Limb Assessment, is a commonly used tool that determines the risk of musculoskeletal disorders based on upper limb posture. While RULA is effective for tasks involving the upper body, like VDT tasks, it has limitations when it comes to assessing overall body posture. Other posture assessment methods, like Quick Exposure Check (QEC), Strain Index (SI), and Posture, Activity, Tools, and Handling (PATH), are also widely used. However, these methods rely on subjective opinions from both the worker and the assessor, leading to inconsistent results and a lack of objectivity. To address this, some studies use 2D imaging or IMU(Inertial Measurement Unit) motion sensors to objectively assess worker posture. Motion sensors, in particular, offer more accurate analysis of worker posture compared to camera footage. Nevertheless, these methods also have limitations, and there is a need for more specific criteria and new risk posture assessment models tailored to the power energy sector. The Heinrich's domino model is a theory that explains the sequential process of industrial accidents. It proposes that accidents occur in five steps. Industrial accidents can be categorized into three causes: 88% are due to unsafe behaviors, 10% are caused by unsafe conditions, and 2% are force majeure incidents. To understand risky behaviors and posture, we investigated various types of accidents in the electric power sector. Based on this research, we developed a model for evaluating risky behaviors and postures in virtual training situations. This model categorizes risk behaviors into three types: electric shock, fall, and inattention. The evaluation criteria for electric shock risk behavior are based on the approach limit distance of the charging part, as defined by KOSHA GUIDE-167-2017. The approach limit distance is classified according to the line voltage of the charging line, and the risk factors are categorized into three levels: safety, warning, and electric shock. For instance, for extra-high voltage lines (22.9kV), the approach limit distance is classified as 183 cm or more. The evaluation criteria for assessing fall risk behavior involve evaluating a worker's likelihood of falling by considering their behavior of extending their body outside the bucket. This evaluation takes into account various factors including the worker's position within the bucket and external elements. To assess the center of gravity of the worker's body, the position of their upper body and the distance to the bottom of the bucket are considered. The assessment relies on body coordinates obtained from motion sensors and calculations of relative mass. Inattentive behavior is defined as a lack of focus on tasks during work and a lack of attention to movement during physical activity. Task inattention is measured by analyzing gaze discrepancy with the task direction, using the Orientation angle measurement. Movement inattention is measured by observing changes in the body's center of gravity and gaze direction. A threshold is set at a change of more than 30 degrees, maintained for more than 3 seconds. Additionally, the risk posture evaluation model is crucial when working on power distribution in the field. The use of insulating sticks is essential to prevent accidents. However, it is important to be cautious of musculoskeletal diseases that can result from using insulating sticks, as they can cause pain and discomfort due to muscle, tendon, and ligament damage. Therefore, creating a safe working environment by preventing postures that can lead to musculoskeletal diseases is necessary. To create a risk posture evaluation model, we use MVN sensors to gather baseline data and measure joints. This allows us to analyze risk posture levels and muscles. Joint angles and muscle activity are factors that affect the occurrence of musculoskeletal disorders. Hence, we employ the Anybody Modeling System, an objective human body model simulation tool, to analyze them. Simulation data is then analyzed using key joint measurement angles, and the musculoskeletal system is simulated based on this information. We collect specific behavior related motion data to observe changes in spinal joint responses and muscle activity. In order to classify the risk level of the upper body and arms, we analyze major joints and muscles and establish criteria for evaluating risky postures. The musculoskeletal evaluation score is determined through the analysis of each joint, such as the upper arm and forearm, and the workload level is calculated by summing the evaluation scores for the upper arm and forearm. The total load score represents the weight of the insulating stick. Risk posture evaluation criteria are categorized into four stages: safety, warning, high, and dangerous, based on the score. For the design and testing of these evaluation models, we utilized our own IMU sensors and Xsens MVN to collect motion data. The data collected was used for risk behavior analysis, posture analysis, and training. They served as the foundation for constructing this evaluation model, including the simulation of joint angles, muscle activity, and more. We also integrated the finalized evaluation model into the system and conducted experiments to confirm its accuracy. Moving forward, our plan is to construct a virtual training system in KEPCO academy and pilot this evaluation system to gather data, refine the evaluation model, and advance it into a technology capable of predicting the risky behavior and posture of workers in the real field.