Machine-readable information models simplify universal data access
Classical communication technologies are mainly used to control the basic functions of machines and plants. In recent years, however, there has been an increasing need for a second channel for new value-added services, such as extended asset management, tools for optimization and diagnosis, the integration of AI algorithms, or for energy management. This shows that the current automation pyramid works well in itself, but is too rigid for new applications. Especially in cases where, in addition to the data used in the processing of basic automation (regulation, control, condition monitoring), applications are being used that improve the performance or value of the plants, machines, components and devices.
For this open world, access via the PLC or the control system is no longer sufficient. The background is that one cannot or does not want to change or expand the existing programminges in the controllers for all the data required. Moreover, in view of the variety and quantity of devices, it is not easy to make data access and configuration simple. The call for machine-readable information and thus for information models is therefore becoming louder. These structure large amounts of data and assign a further description to a variable. After all, the subsequent added value lies in the algorithms and not in the data transport.
Users of PI technologies are already familiar with this in the form of feature lists and profiles. Now it is a question of making these information models machine-readable, relating existing information models to each other and lifting them into widespread use. This provides the user with a practical and, above all, universal solution.
Information models with PROFINET
Uniform semantics as key to machine-readable data
In general, the trend towards information models is not new for users of PI technologies. If you look at the various PI application profiles, for example PROFIBUS PA, PROFIdrive or PROFIenergy, these have made the daily work of users much easier in recent years, especially with regard to device interfaces.
In the future, however, it will be necessary to follow this path via a machine-readable information model. The mental work for this has already been done thanks to the application profiles that have proven themselves many times over. Standardized OPC UA interfaces also simplify application configuration. This means that the user not only receives values of the variables, but also the description of the structures and attributes. Since not all devices support PI application profiles and not all are connected via PROFINET, further concepts are needed to obtain machine-readable information of the variables.
The advantages are that this provides very easy access to the data and thus saves a lot of time when integrating new components. Thanks to uniform semantics, the quality of the data also increases. Since the amount of data from the field will certainly continue to grow, scaling is made easier with such approaches. And ultimately, these information models pave the way for future technologies.
Assigning information clearly
Structuring information according to its allocation to different task areas is helpful. These include the physical view (devices, components, etc.), the functional view (functions, contexts, applications, etc.) and the communication view (communication system, alarms, addresses, data transport, etc.).
In general, neither the origin nor the transport route of the data is decisive for application algorithms. Depending on the assignment to their view, the information of the automation system or instrumentation is described along the access path. All required information is thus available at the interface to the application. Only the information model makes the relationship between the technological PI portfolio transparent to the user, in the way that all information can be clearly assigned to each other, and the origin as well as the access path is clear.
A look at the details
The great variety of automation technology devices, the different PI technologies and the different use cases in the life cycle of factories make further categorization necessary. For practical application, different device types, function groups andtechnologies are therefore separated from each other at their interfaces. This creates a modular structure that can be combined with one another depending on the application requirement and use case. The elements of this refinement level are called facets. This results in individual models that represent the three named views and collectively describe an automation entity.
|1.||Base facet||The Base Facets are dedicated to the physical aspects of the automation devices and the communication system.|
|2.||Functional facet||The Functional Facets form the core of the information models, in which technology-independent function-related data are described from the application's point of view. The data are structured according to individual functions (e.g. measured temperature value) or function groups (e.g. energy management).|
|3.||Composition facet||The composition facet (Composition/Companion Specification) combines the functional facets into specifications that can be implemented.|
PI is continuously working on further upgrading its own technologies for Industrie 4.0 applications, especially with regard to sensor-to-cloud connectivity. After all, the approximately 35 million devices in the field that have a PROFINET interface open up access to valuable data.
Currently, the development of standardized information models is a major focus at PI. For example, numerous facets of the PI information models already exist in relation to pressure, temperature, status, but also more general ones for energy management or the health status of sensors, for example.
Further mappings in OPC UA are currently being developed for PROFINET in order to use device and diagnostic data in IT applications without much effort. It is particularly important to design the interaction between PROFINET and the OPC device integration modelsin such a way that it is easy for the user to get started. Only in this way will information models find their way into widespread use.
Current information models in action
Typical use cases that benefit from the use of information models are:
- Clear identification of all equipment and components in manufacturing plants.
- Comparison between the planning documents and the actually constructed system
- Equipment condition monitoring
- Monitoring the condition of the plant (condition monitoring)
- Life cycle tracking of the equipment
- Digital shift log
- Analysis of the correct dimensioning of the installed units and components
- Plant-wide recording of the firmware and hardware versions of a unit used for a central update.
Some use cases are already being implemented, as this example shows: PROFINET I&M data includes firmware, hardware version, serial number, device names, among others. This proven data has been mapped to OPC UA in a Companion Specification. The different views are challenging. For example, from a technological point of view, production data or process sequences of a machine are important, whereas from a maintenance point of view, the devices used are interesting. The current work is therefore about making the different views of the same machine compatible and referenceable.
Another example comes from the IO-Link community: a 'non-functioning' gripper quickly becomes noticeable. However, it would be much more interesting to get a statement on the basis of the gripping pressure as to whether the force is still sufficient after many cycles or whether a check or maintenance is necessary ahead of time. The same applies to the degree of contamination in sensors. In principle, many sensors already provide such data today. This diagnostic data would have to be delivered to the cloud or another location, virtually bypassing the PLC, so that intervention can take place at an early stage - and this without having to laboriously rebuild the automation solution. The basis for this can only be open sensor-to-cloud communication. For this purpose, the IO-Link community, in cooperation with the OPC Foundation, has defined the Companion Specification "OPC UA for IO-Link" and the use of the JSON exchange format for IO-Link.
Process Automation Device Information Model (PA-DIM), which was jointly developed by ZVEI, FieldComm Group, the OPC Foundation and PI, is also an OPC UA information model. This allows software applications to access device information in the process industry without additional mapping. The NAMUR Recommendation NE 175 and NE 176 gives a well defined set of use cases such as “Automated as built”, “Unique Identification” and “Multivariable possibility check”.
Connection of plant and device information models
Localization of plants via connected information models
Here you can find several frequently asked questions:
In general, there are three possibilities for obtaining data with advantages and disadvantages.
Data access from the PLC
Data access via edge devices
Direct data access in the PROFINET device
First, check whether the control, edge device or PROFINET device has an OPC UA interface. If this is the case, it must be examined whether the OPC UA interface supports a known adopted OPC UA Companion Specification. Then it is possible to work with the objects and variables defined in it. If this is not the case, the information model of the OPC UA interface must be examined individually for the objects and variables suitable for the application.
There is the "OPC UA for PROFINET Companion Specification" and the "Joint PI and OPC Foundation Working Group Displays PROFINET in OPC UA".
There is also the "OPC UA for IO-Link Companion Specification".
Moreover, the "OPC UA for Energy Management" is currently under PI review.