January 31, 2024
In the fast-paced world of industrial production, efficiency and productivity are critical factors for success. In the building technology industry, the utilization of assembly stations plays a vital role in determining overall performance. In this article, we will delve into the concept of station utilization, its significance in the industry, and explore our innovative system for measuring it digitally.
Station utilization refers to the degree to which an assembly station is being effectively used over a given period. It is typically measured by the percentage of time a station is occupied by a worker, machine, or process compared to its total available time. A high station utilization implies that the assembly station is efficiently and optimally being utilized, contributing to enhanced productivity and reduced idle time.
Station utilization is a crucial performance indicator in the building technology industry for several reasons:
a. Resource Optimization: By monitoring the utilization of assembly stations, manufacturers can identify bottlenecks and inefficiencies in the production process. This data enables them to optimize resource allocation and ensure that each station contributes to the overall workflow seamlessly.
b. Production Planning: Efficient station utilization data aids in better production planning. Manufacturers can schedule shifts, breaks, and maintenance activities based on the real-time utilization data, resulting in a smoother production flow and higher output.
c. Cost Reduction: Identifying underutilized stations allows manufacturers to avoid unnecessary expenses associated with idle time and wasted resources. Optimizing station utilization leads to cost reduction and increased profitability.
d. Performance Evaluation: Station utilization serves as a performance metric for evaluating the effectiveness of different production lines or workstations. It helps in making informed decisions regarding process improvement and resource investments.
In this process, a straightway valve is taken from an intralogistics delivery box and is packed in a box together with a valve adapter and operating instructions, taped up, labeled, and stowed in another intralogistics box.
Our system employs AI-video-analytics technology to automatically measure station utilization in real-time. Using a combination of a 2D camera system, data analytics, and edge computing, our solution provides valuable insights into production line efficiency.
a. Hardware Integration: The camera system is placed on top of the station to recognize the different hand movements of the operator. The camera collects video data and sends it into the small edge device which is installed on the lower part of the station, where the data is processed.
b. Data Analysis: The data collected from the camera is processed and analyzed using pretrained deep neural networks to detect hand coordinates with a landmark detection approach and logical elements like movement zones to separate different hand movements of the operator. The system calculates the percentage of time each station is active and compares it against its total available time.
c. Real-time Monitoring: Our system offers real-time monitoring capabilities, enabling production managers to track station utilization on dashboards. This feature allows for immediate adjustments and interventions in case of suboptimal utilization.
In this example, although the target time of 35.2 seconds is met and production is therefore at sufficient capacity, it is also clear that the packaging station(station 3) with the highest cycle time of 28.1 seconds in the median is the bottleneck. This can become a problem if demand is increasing. A look at the utilization of the station also shows that station 3, with 30% downtime, still has potential for optimization. It is therefore worth taking appropriate countermeasures here. In this way, various problem areas in production can be prioritized and bottlenecks in production can be counteracted at an early stage. As the data is available in real time, the effect of the measure can also be checked immediately after it has been taken.
Supported by Intel, we are proud to be bringing this solution to market. The solution is optimized for data-intensive workloads and is adaptable, vetted, and ready for immediate deployment.
The solution runs on Intel® Core™ i5 Processors which are state-of-the-art processors allowing for maximum flexibility and performance in Industrial Settings. Best in class Wi-Fi connectivity with Intel® Wi-Fi 6 (Gig +) ensures a responsive and reliable connection for immersive connectivity even in large factory spaces.
To create training data to pretrain a model for advanced hand-recognition even with working gloves, we use the Intel® RealSense™ Depth Camera D435f. With its advanced depth-sensing capabilities, this camera is designed to capture accurate 3D images and video in real-time, making it ideal for our industrial use case.
With the Intel® OpenVino™ Toolkit, we optimize our AI models to run efficiently on Intel hardware, unlocking unparalleled performance and accuracy. A lot of workstations need to be equipped, OpenVino™ makes it easy to deploy our AI models at scale.