Different sensors, a gateway and task-related information retrieval algorithms are required for a data retrofit.

© SSV Software Systems

Technically, a data retrofit solution should be as independent as possible from the machine control system (PLC) and the sensors already installed in the machine and connected to the PLC. In other words, intervention in the time-critical 'control loop' of sensors, PLC and actuators should be avoided as far as possible. This also minimizes IT security risks. A data retrofit therefore requires task-related sensors, a suitable Industry 4.0/IIoT gateway plus the software components for obtaining information.

A suitable calibration and metrological procedure is embedded in an algorithm directly in the gateway for each sensor. In addition, the gateway makes the data available for transfer to a cloud.

© SSV Software Systems

Before selecting the sensors, however, it must be determined as far as possible what information is ultimately required. The market offers a large number of suitable sensors for almost every measurand. However, in order to obtain valuable information from the raw sensor data, numerous intermediate steps are required. First of all, this includes the selection of a suitable metrological process, including calibration for each individual sensor, in order to generate the appropriate data for the measured variable. This requires extensive specialist knowledge.

If the measured variable-related data is available in the gateway, the required on-site information is obtained using a suitable algorithm. Appropriate data analysis know-how and expert knowledge of the respective mechatronic assemblies is required for the design and implementation of such an information acquisition algorithm. It must be taken into account that different sensor data of a machine are linked together to obtain information (on-site sensor data fusion). For example, in order to clearly distinguish between machine damage and overload, the algorithm may have to link the data from a current and voltage measurement with the output of an imaging sensor for object detection.

Who needs data?

Most mechanical engineering companies have complex value chains with different levels and instances. Within these structures, there are numerous interest groups for whose daily work the product usage data and the information derived from it would be of considerable value. Here are three examples:

1. Service: The main task of a service team is to ensure optimum customer satisfaction for the entire product service life. This goal can be achieved primarily through effective support in avoiding unplanned machine downtimes, fast response times when service is required and the shortest possible delivery times for spare parts. Timely information on usage intensity (operating hours, capacity utilization), current ambient conditions (pressure, voltage, frequency, temperature), information on vibration behaviour as well as error and fault messages help the service management.
2. Marketing and sales: Detailed specifications for the further development of existing products are expected from marketing and the associated product management. Experience has shown that this requires, among other things, usage information regarding the individual product features (question: Which product features are used how often?). A sales department should be provided with suitable information on whether the customer could use any enhancements and when it makes sense to offer a successor to the machine currently in use. Proactive service products can also be sold with appropriate usage data and the information that can be derived from it.
3. Technology development: Development is responsible for ensuring that the technical data of a machine is actually delivered to the customer in practical use. In this respect, there is great interest in operating data and information that shows that the features sold are actually available. In addition, information on ambient conditions, vibration behavior as well as error and fault messages are also of considerable importance for employees in development.

Added value on the customer side

Similar value chains also exist on the customer side. In addition to the requirement from production management that machines should have standards-based data interfaces in order to be integrated into a multi-vendor production environment or system, there is also an increased interest in machine data for information acquisition. Two examples of possible interest groups:

• Maintenance: Operational maintenance carries out maintenance work, inspections, optimizations and other work to maintain the functional condition of all machines and systems and to restore them as quickly as possible in the event of a failure. In addition, unplanned downtimes are to be avoided by means of preventive maintenance. With the help of permanent machine usage data, all maintenance tasks can be coordinated much better and organized more effectively.
• Controlling: Behind every machine that a company purchases, there is a special investment calculation for a usage period of several years. This calculation takes into account investment and operating costs as well as the value of the goods produced using a machine. In addition to amortization, the aim is to achieve a return on the capital invested. An investment calculation therefore includes numerous assumptions regarding production output (number of units produced, number of defective parts) or energy requirements (such as quantities of auxiliary and operating materials). Using current machine data, financial controlling can use a gap analysis to monitor in real time whether the amortization and return expectations will be met and initiate corrections in good time in the event of deviations.

Dual communication skills

The gateway module of a data retrofit solution requires a configurable OPC UA interface with a suitable information model in order to pass on the on-site information to the machine operator's various interest groups. This interface should also be supported by a machine data app model in order to be able to respond to changes in data user requirements with an app that can be installed at a later date. In the IT environment, a behavior known as continuous delivery has already been established for similar tasks, which ensures permanent enhancements or error corrections based on changing customer requirements and test results.

An anomaly detection algorithm can be used in the gateway. This compares the sensor data with a predefined set of rules generated by a cloud service.

© SSV Software Systems

In order to provide the machine manufacturer as the provider of a retrofit solution with the desired status and product usage data, the gateway must send the measured variable-related data from the sensors in time series data format to a cloud service via MQTT at certain intervals. The machine operator's consent is required for this. As a valuable additional benefit, the machine manufacturer can offer its customers anomaly detection based on the latest data analysis techniques in return. To this end, the time series data stored in the database is evaluated from time to time using descriptive and explorative analysis methods. The information retrieval algorithms used in this process provide the customer gateways with an increasingly precise set of rules for recognizing assembly and system anomalies. A continuous delivery service should also exist for these algorithms. Finally, the time series stored in the database can also be used to forecast trends for predictive maintenance applications, future energy and resource costs and service dates with the help of appropriate extensions, which in turn enable cost optimization.

Author:
Klaus-Dieter Walter is a member of the management board at SSV Software Systems.