In which product areas are you already using AI?

Phoenix Contact deploys AI comprehensively across all product areas of the company – from product development and product management to production, logistics, and customer-facing applications. Specifically, AI solutions are found in the following areas:

• Automation: Analysis of data captured via the I/O system and processed in the controller. For this, we use MLnext, a solution for generating and executing machine learning. It can also be scaled to edge PCs and our IPC/PPC portfolio.
• Building automation: We offer cloud-based control of systems and components to optimize building operations using AI and self-learning algorithms.
• Sector coupling: AI can be used as a cloud service to forecast energy consumption and generation. This enables smarter decisions for controlling adjustable consumers, such as charging stations, and energy storage, thereby improving energy flows.

How are AI-supported systems changing your automation solutions?

Automation here plays a dual role. On the one hand, model predictions can be used as decision inputs for automation – for example, to sort out defective parts or trigger alarms when atypical behavior occurs. On the other hand, automation is the interface to the physical world. It must ensure that people and machines are protected from AI decisions. That means preventing data manipulation and avoiding the collection of personal data.

In this environment, AI can also be used for code generation. Software functions and architectures can be pre-generated, with the programmer only needing to verify correctness. AI can also take over highly repetitive tasks completely, such as testing or documentation. Moreover, it enables data-based rather than purely rule-based decisions. AI can analyze complex diagnostic situations, classify error patterns, and provide decision support for troubleshooting. This expands the diagnostic functions of a controller.

In addition, AI supports virtualization. Since it is trained and used at the edge or IT layer, automation also moves into these layers. In combination with our virtual PLCnext Control, AI and automation systems can be scaled together from the field level all the way to the cloud.

What role does AI play in your IoT platform?

Through the ecosystem approach of PLCnext Technology, AI solutions can be loaded onto PLCnext Control like an app – to collect, store, and visualize data. AI solutions are provided both by Phoenix Contact and by third parties, who transfer their domain expertise into AI apps.

Thanks to the openness of PLCnext Technology, solutions can be deployed virtually as containers or software libraries. This builds a distributed IoT system where AI is used exactly where it is needed: directly on the PLC for short response times, in the edge/IT layer for analyzing entire plants or lines, or in the cloud. AI thus becomes another tool in the toolbox of the PLC programmer – and enables machine builders to implement digital business models.

Can you share a practical example?

For instance, a press plant in the automotive industry is monitored by AI installed on the PLCnext Edge PC. This makes it possible to detect gradual damage to an electric motor at an early stage. Maintenance can then be carried out before the system fails. By reducing unplanned downtime through predictive maintenance, system availability increases.

As another example, a plant manufacturer uses AI models to classify raw material. The time between data acquisition and model prediction must be no more than 200 ms. For this, the manufacturer uses a PLCnext Control with integrated AI and OPC UA as the communication protocol. For customers, AI integration is a clear added value, since raw material no longer has to be classified manually. This improves both performance and autonomy of the plant.

How do you ensure data quality and security?

PLCnext Technology sets a high standard here. Our advantage is that data quality for processing by a PLC must inherently be very high. Phoenix Contact’s entire portfolio for signal duplication, interface and I/O modules, and controllers is therefore designed for reliability in data acquisition and transmission.

Thanks to our long-standing experience and involvement in fieldbus standards – such as the current TSN – we place great emphasis on timing accuracy and synchronization. Every data point must be traceable to its origin, context, and timestamp.

The recorded data always reflects the behavior of the plant and processes, and thus includes sensitive information that must be protected against unwanted external access. This means that the entire infrastructure must be robust against cyberattacks. By implementing the cyber security standard IEC 62443, Phoenix Contact ensures these requirements are met in its products and development processes. In addition, we act as a consulting partner for the NIS 2.0 directive, the Cyber Resilience Act (CRA), and the Machinery Regulation (MVO) to offer customers a holistic cyber security approach.

To scale AI-based data analysis, information standards for components and systems must be developed. Since many AI systems are still under development or evolving rapidly, integration in OT must also be rethought. Permanent updates for field devices and PLC systems, as well as their applications, must be implemented easily but securely. For this, device and update management is essential – and is a core part of Phoenix Contact’s offering.

How do AI solutions support energy efficiency and sustainability?

When media consumption is recorded across a site, AI can automatically detect energy-saving potentials – such as unnecessary standby consumption or energetically unfavorable machine configurations.

It is also essential to collect information about building usage in order to correlate consumption with occupancy. This way, it can be determined when and where supply media must be provided.

Optimal control of adjustable consumption and energy storage requires not only knowledge of current states but also forecasts of future energy generation and consumption. Based on historical data, AI can predict these loads for each site, identifying consumption patterns or unplanned peaks.

Sustainability and productivity do not conflict – they reinforce each other. Sustainability means producing the maximum number of products with available resources. Therefore, control in terms of availability, productivity, and quality usually leads to higher efficiency and more sustainable use of resources. Phoenix Contact offers solutions for condition-based maintenance and automatic quality control. To reduce setup times, the digital nameplate can be used to preselect stored configurations.

How do you involve partners and customers in AI projects?

With the PLCnext Technology ecosystem, we provide a platform on which PLCnext partners can make their software solutions for data analysis available. Through the PLCnext Store, we bring together AI users and corresponding solution providers.

As a platform, PLCnext Technology is designed for powerful and secure data processing for training and using AI. The approach can be scaled from the controller to the edge layer and into Proficloud.io. High flexibility can also be achieved with the virtual PLCnext Control, which can be adapted as a platform from IT to the cloud.

Today, AI solution business is mainly project-based, with Phoenix Contact acting as the technology provider. Depending on customer requirements, we draw on the partner network from the PLCnext Store or our portfolio. Often, a system integrator is involved, equipped with our solution portfolio and trained to implement the projects.

Domain experts are supported by our solutions such as MLnext Creation, which enables them to use AI without data science expertise. By lowering the implementation barrier, customers can scale their AI use cases more easily across departments.

More and more machine builders are starting to create their own AI-based services. They use PLCnext Control as the interface for collecting and storing data, as well as the execution environment for AI (MLnext Execution). To ensure developer flexibility, Phoenix Contact relies on an open exchange format for machine learning models (ONNX). Users can either work with the user-friendly web interface or automate processes via REST API.

How do you integrate employees into the daily use of AI?

Our Digital Innovations division offers so-called ideation workshops. These train employees in their respective departments. Several hundred colleagues have already been enabled in full-day workshops to identify and successfully implement their own AI use cases.

We also offer leadership workshops on AI, since managers face particular challenges in using AI. In the departments, AI experts are trained or hired. Phoenix Contact differentiates between AI users and AI developers. Both groups receive basic training in compliance and AI methods. The expert network also exchanges knowledge through lecture series (AI Meet-Up) and workshops (Data Camp).

Where is Industrial AI heading at your company?

Every company will have to tackle the task of integrating AI into its processes and products. Phoenix Contact wants to actively shape this transformation by aligning its portfolio optimally for customer and partner AI use. We achieve this through openness in data collection (MQTT, OPC UA, gRPC) and AI execution (ONNX, Matlab), supported by cyber security (IEC 62443).

Industrial AI will increasingly be embedded into use cases to create added value. A distinction can be made between direct AI integration and the integration of an interface for AI use. AI systems in which, for example, a machine learning model is firmly integrated into the software or service could be applied in an increasing number of product families.

Furthermore, hardware and software should also be configurable via AI agents – such as ChatGPT or MS Copilot. A device, for instance, could be configured via a prompt given to the AI agent: “Set the IP address, apply our global firewall policies, perform a firmware update, and then deploy the application to the device.”

We are currently working on a framework for agents that, for example, secure the use of these interfaces. A company-wide governance structure will be in place for this.