InGrid

For the Danube InGrid project we designed and implemented a solution for the near-real-time, cross-border exchange of meteorological measurements and warnings between the Slovak distribution grid (ZSD) and the Hungarian side (E.ON / EED). Both sides operate their own, mutually independent data platforms. Our goal was to connect them so that they could exchange weather-station data reliably, securely and in a standardized format — without either side depending on the availability or technology of the other.

The brief

High-voltage power lines in the Slovak-Hungarian border region are exposed to the same weather risks — icing, strong wind and frost — which do not respect national borders. Yet each operator only monitors its own meteorological stations.

The operators therefore faced the task of:

• exchanging measured data and warnings about dangerous phenomena every minute,
• connecting two technologically distinct and independent platforms without creating a hard mutual dependency,
• while ensuring a high security standard and a unified, machine-processable message format.

At the same time, the solution had to be ready for future growth — a larger data volume and a higher transfer frequency.

The solution

We chose Azure Event Grid as the shared integration element, serving as a shared “exchange” layer for transferring messages between the two sides. A message is either a warning or a measured value.

Key design principles:

• Unified message format — all messages are JSON-encoded according to the CloudEvents 1.0.2 specification, with the same structure and descriptive elements in English. As a result, every warning is self-describing: it contains the rule that triggered it, the parameters used, a multilingual description and the device’s measured values. From such a message you can directly create, for example, an SMS, a push notification or an e-mail.
• Platform independence — both the Slovak and Hungarian sides process received data on their own. The SK side stores it in a relational database for later analysis; the HU side stores it in a Data Lake for machine learning and displays the warnings in reports.
• Secure connection — communication runs over a site-to-site VPN between the E.ON network and the Slovak side, with authentication when publishing to Event Grid.
• Readiness for scaling — in case of a significantly larger volume or higher data frequency, the solution is designed so that it can move to Azure Event Hub with a longer retention period and support for batch transfer.

Communication is organized into topics and subscriptions — separately for warnings and separately for measured data of each country. A subscriber can thus filter exactly what they need: a specific country, selected measuring stations or only a certain warning type (for example, devices near the Slovak-Hungarian border).

Architecture of the Slovak side

On the Slovak side we built a processing chain running in the OpenShift container environment:

• Data collection from the SCADA system and from SHMÚ data using dedicated reader services (Java / Spring Boot).
• Event streaming via Apache Kafka, which collects measured data and warnings into topics.
• InGrid Data Publisher, which publishes warnings and measurements as CloudEvents messages to the Hungarian side’s webhook, and InGrid Data Consumer, which receives data in the opposite direction.
• Storage and analytics — data is stored in a PostgreSQL relational database and is available to tools for evaluating measurements and warnings.

On the Hungarian side, the solution feeds into a Databricks pipeline for processing warnings, a Data Lake and Power BI reports.

How it works in practice

When a measured value satisfies one of the defined warning rules, the system generates a warning with all related measured data and sends it every minute for as long as the rule applies.

Measurements of all key quantities are supported — global radiation, air temperature and pressure, relative humidity, wind direction and speed, precipitation and icing (measured on the Slovak side).

The warning rules cover real operational risks, for example:

• wind speed in graded levels (strengthening, strong and dangerous wind),
• air temperature (frost and dangerously low temperatures),
• icing formation on overhead line conductors (a combination of below-freezing temperature and a negative dew point),
• on the Slovak side, also exceeding the rate or weight of ice accretion and the maximum wind speed.

Each warning carries a unique type code, severity level, rule validity and information about who last changed the rule and when — which simplifies any later auditing.

Benefits

• Early warning of icing and strong wind on cross-border lines, and thereby greater resilience and reliability of the distribution grid.
• Reliable transfer with a delivery guarantee (at-least-once), retries and dead-letter handling of undelivered messages.
• Standardization and extensibility thanks to the open CloudEvents specification and unique identifiers (ULID) — the solution is ready for adding more stations, rules and partners.
• Preserved autonomy of both sides — neither platform depends on the operation of the other.
• Future-readiness — a smooth transition to Event Hub as the data volume grows.

Conclusion

The solution shows how two independent energy platforms can be connected into a single smooth flow of data — securely, in a standardized way and with room for further growth. The project is part of the Danube InGrid initiative and is co-financed by the European Union.