Introduction
Modern technologies allow production in Ukraine to implement digital twin and artificial intelligence (AI) to predict component fatigue. This ensures high reliability, reduced downtime and cost optimization. Choosing the right solutions supported by DSTU, EN, ISO standards is becoming critical for manufacturing enterprises.
How it works
The digital valid component is created on the basis of data from sensors that record temperature, pressure, rotation frequency and other parameters. AI algorithms analyze this data, identifying patterns that indicate damage or fatigue. This allows you to predict the moment when the component will no longer meet the technical requirements, before the failure occurs.
Data requirements
A minimum of 12 months of sensor data collection is required for AI to work effectively. This volume provides statistical significance for building the model. The number of sensors should be between 5 and 20 per component, depending on the complexity. Sensors must comply with EN 60751 (for temperature sensors) and ISO 9001 (for data quality) standards.
Required data types:
- Temperature (°C): measured in the range of -20 to 150 °C
- Pressure (bar): measured in the range of 0 to 100 bar
- Rotational frequency (rpm): measured with an accuracy of 0.1%
- Vibration (μm/s²): measured with an accuracy of 0.5%
- Electrical parameters (kV, A): measured with an accuracy of 1%
Implementation architecture
The system includes several stages: sensors, edge servers, accounting server, and analysis-based actions.
1. Sensors
Sensors mounted on components collect data in real time. They comply with EN 60751 and ISO 9001. standards Data collection occurs at a frequency of 10 Hz.
2. Edge servers
The edge servers process the data in real time, filter the noise and transmit the results to the accounting server. Processing is performed using machine learning algorithms that comply with ISO/IEC 23894. standards
3. Accounting server
The accounting server stores the data, creates the digital valid component and performs the prediction. Models conforming to the ISO 23247. standard are used
4. Actions
Based on the forecasts, actions are taken: replacing components, scheduling maintenance and sending messages to production managers.
Real results
The following results were obtained as a result of the implementation of the component fatigue forecasting system in Ukrainian factories:
1. Reduction of downtime
On average, downtime is reduced by 35% due to the ability of the system to detect unwanted changes in the operation of components in a timely manner.
2. Return on investment (ROI)
Implementation of the system requires an investment of 15,000 to 30,000 €. Return of the investment is achieved after 12-18 months.
3. Increased durability
Durability of components increases by 20-30%, which meets ISO 13374. standards
Limitations and challenges
The introduction of AI systems has a number of limitations:
1. Dependence on data quality
If the sensors measure incorrectly or lack data, the model cannot work effectively. It is necessary to ensure the quality of data and their volume.
2. High cost of implementation
Implementation of the system requires significant costs for sensors, servers and support. Not all businesses can afford such costs.
3. Technical complexity
Implementation of the system requires coordination between IT and OT, which can cause complications. It is necessary to perform technical preparation and training.
Building your own system or purchasing a ready-made solution
The choice between building your own system and purchasing a ready-made solution depends on specific needs.
1. Building your own system
Building your own system requires technical knowledge, significant resources and time. The cost can range from €50,000 to €100,000. Using ready-made solutions reduces risks.
2. Purchase of a ready-made solution
Off-the-shelf solutions usually cost less and have support. The cost can be from 20,000 to 50,000 €. They meet ISO 23894 and EN 60751. standards
Start of work
To get started, the production engineer can perform the following steps:
- Assess the current situation with production and maintenance.
- Identify the components that require implementation.
- Install sensors according to EN 60751 and ISO 9001. standards
- Choose the system architecture: edge or account server.
- Perform data preparation and model setup.
- Carry out testing and implementation.
- Conduct training for IT and OT teams.
Conclusion
The introduction of a digital valid component and artificial intelligence allows for a significant improvement in the production process. Choosing the right solutions supported by DSTU, EN, ISO standards is critical to success. Unitex-D provides a wide range of components and services that support this digital transformation.
List of used sources
- ISO 23894:2019 - Models of the digital valid component
- ISO 13374:2013 – Status monitoring
- ISO 9001:2015 – Quality management systems
- EN 60751:2016 – Temperature sensors
- EN 60751:2016 – Temperature sensors
- DSTU 3018-2007 – Technical conditions for products
- ISO/IEC 23894:2019 - Models of the digital valid component
- ISO 23247:2017 - Standards for the digital valid component
