Industrial edge computing places data collection and processing between shop-floor equipment and enterprise IT systems. For automation engineers and operations leaders, this middle layer can support faster use of production data, coordination with business systems and feedback to machines. Mitsubishi Electric’s MELIPC industrial computer series illustrates how edge computing can be combined with real-time device control in manufacturing environments.
What industrial edge computing does on the factory floor
Production equipment generates data that can inform monitoring, analysis and operational decisions. Sending every item of raw data directly to an IT system is not always the only useful approach. An edge layer can collect shop-floor information, process it locally and pass relevant results to applications or higher-level systems.
This creates a practical connection between operational technology and IT. Edge computing can support functions such as:
- Collecting production and facility data from automation equipment.
- Processing and organizing data close to the source.
- Providing information to monitoring, analysis and diagnosis applications.
- Returning appropriate results or feedback to the shop floor.
- Coordinating production information with enterprise or cloud-based systems.
The objective is not simply to add another computer to a control cabinet. It is to make production information more usable across the plant while supporting operational needs such as real-time response, quality improvement and maintenance planning.
Combining edge processing with real-time industrial control
Some applications require both information processing and deterministic control. These have traditionally been implemented with separate computing and dedicated control devices. The MELIPC MI5000 is designed to run Windows and the VxWorks real-time operating system on one module, assigning different work to each environment.
VxWorks is used for device control and data collection where deterministic performance is required. Windows can run applications that display or process collected information. Data can pass between the two operating systems through shared memory and shared folders. The product information also describes the ability to restart Windows independently while VxWorks continues running, separating recovery of the information-processing environment from ongoing real-time operation.
This approach can reduce the need for separate devices in systems that require both control and information processing. It does not remove the need to assess application timing, network design and machine-level safety requirements; those remain part of system engineering and validation.
Choosing an edge computer for the application
Edge computing requirements vary across production lines. A system integrator may need real-time control, while another installation may prioritize operator visualization, equipment connectivity or a compact footprint. MELIPC models address a range of these roles:
- MI5000: Combines Windows and VxWorks for applications involving real-time control and information processing. Its listed network interfaces include CC-Link IE Field Network and CC-Link IE Field Network Basic.
- MI3000: Combines an industrial computer with an integrated touch display. Pre-installed Edgecross Basic Software, an SLMP Data Collector and GT SoftGOT2000 support shop-floor data collection and visualization.
- MI2000: Provides computer-based analysis and monitoring, with expansion and storage options described for applications requiring additional capacity or interfaces.
- MI1000: Offers a compact, fan-less computer option for installations where available space is limited.
These distinctions are useful when defining system architecture, but model selection should also account for the required protocols, connected devices, software, environmental conditions and expansion needs.
Connecting factory automation and IT systems
Open edge platforms are intended to help coordinate data from automation equipment with higher-level systems. In the MELIPC context, Edgecross provides functions for real-time data processing, data-model management and gateway communications. Data collectors connect to shop-floor networks and devices, while edge applications perform tasks such as monitoring, analysis or diagnosis.
Data-model management represents devices, machines and production lines in a hierarchical structure. This can make it easier to organize and maintain production data across equipment that might otherwise be managed separately. Gateway communications support information exchange with IT systems, including cloud environments, using protocols such as MQTT, as described in the platform material.
Connectivity depends on the selected hardware and software. The MELIPC range lists support options including SLMP and OPC UA data collectors, with other data-collection capabilities varying by model and configuration. Integrators should verify compatibility and any required interface hardware before finalizing a design.
Using production data for diagnosis and improvement
Collected data becomes more valuable when it can inform decisions or trigger a timely response. The Real-time Data Analyzer described for the MELIPC platform supports offline analysis of historical data and real-time diagnosis. Offline analysis can help establish rules from past operating data; those rules can then be applied to incoming data, with diagnosis results returned to the shop floor.
The analysis methods include similar-waveform recognition, the Mahalanobis-Taguchi system and multiple regression analysis. These methods address different questions:
- Similar-waveform recognition compares incoming waveforms with patterns learned from normal operation to identify deviations from typical behavior.
- Mahalanobis-Taguchi analysis considers relationships among multiple measured variables when assessing how far current data differs from a reference distribution.
- Multiple regression analysis estimates a target variable from several measurable inputs, which can be useful when the target cannot be measured directly at that point in the process.
Analysis results depend on the data, the selected method and the diagnosis rules developed for the application. These tools can support maintenance and quality work, but they do not replace sound knowledge of the process or appropriate validation before feedback is used in production.
Practical considerations for deployment
A successful edge architecture starts with a clear operational need. Define whether the priority is real-time control, visualization, data collection, analysis or IT coordination. Then map the devices and networks involved, identify which data needs local processing, and determine how results should be shared with operators, control applications or enterprise systems.
Also consider how the system will be programmed, monitored and maintained. The MELIPC materials describe tools for configuration, event-history diagnostics and application development, as well as solution templates that can provide a starting point for certain applications. Templates can be customized, but compatibility, integration effort and the plant’s own requirements still need to be assessed.
Conclusion
Industrial edge computing provides a way to connect production equipment with data processing and IT coordination closer to the factory floor. When designed around the application, it can bring together equipment data, analysis, visualization and—in suitable systems—real-time control and feedback. For plant teams, the key is to select an architecture that fits the control task, connectivity requirements and intended use of production data, then validate how it will operate within the wider automation system.