DYNVOLT Solar SCADA and Forecasting Platform Presented at EU PVSEC 2026 With Only 5.1% Day-Ahead Forecast Error

Zoran Aleksikj presents DYNVOLT on the Startup Solar Launch Pad stage at EU PVSEC 2026, Rotterdam.

Co-founder and CEO Zoran Aleksikj presents day-ahead PV forecasting results verified on live plants, with methodology and exclusion ledger published

ROTTERDAM, NETHERLANDS, September 16, 2026 /EINPresswire.com/ -- Zoran Aleksikj, co-founder of DYNVOLT and CEO of Dynode.ai, presented the DYNVOLT solar SCADA, energy management (EMS) and energy market intelligence platform at EU PVSEC 2026, the 43rd edition of the conference, in Rotterdam (14 to 18 September). DYNVOLT exhibited in the Startup Pavilion, and Aleksikj pitched on the Startup Solar Launch Pad stage, presenting verified results for the platform's new AI production forecasting model.

DYNVOLT is an edge-first platform for utility-scale solar and storage that connects the physical plant to the electricity market in one system. Every reading from every device flows into one normalized data model, and that same clean, validated data feeds string-level SCADA, AI production forecasting, day-ahead and intraday bid construction, battery dispatch optimization, and O&M and reporting. DYNVOLT is in production at utility-scale plants in North Macedonia, monitoring 840+ PV strings at 30-second resolution and issuing the day-ahead forecasts used to nominate that production on the power exchange.

Verified day-ahead solar forecasting results

The headline of the presentation was DYNVOLT forecasting model v2, live since September 2026. In a walk-forward out-of-sample evaluation covering April to August 2026, it achieved a normalized mean absolute error (nMAE) of only 5.1% of AC capacity (range 5.0 to 5.2%) on day-ahead forecasts during daytime hours, with normalized RMSE of 8.2 to 8.7%. The best published academic result for comparable plants, converted to the same convention, is about 8.6% nMAE; smart persistence on DYNVOLT's own plants scores 12.8%.

The model blends ten numerical weather prediction sources at their true day-ahead issue time, applies per-plant machine learning correction trained on each site's own history, and retrains nightly. Because one data model carries telemetry, curtailment commands and market schedules, the training pipeline recognizes when a plant was ordered not to produce and excludes those hours automatically, something a standalone forecast vendor typically cannot do because it does not see the SCADA command log. P10 to P90 bands are recalibrated daily and reached 80% empirical coverage. DYNVOLT publishes its full methodology and exclusion ledger, listing every day removed for curtailment or telemetry outage, so the figures can be reproduced.

"Most forecast accuracy numbers in this industry are quoted without a definition of what was measured," said Zoran Aleksikj, co-founder of DYNVOLT. "We brought a number to EU PVSEC that a researcher would accept: day-ahead vintage, daytime hours, normalized to capacity, every excluded day listed. 5.1% on live plants, scored against the meter and used to settle real imbalance positions."

Aleksikj added: "The result comes from one data model and clean data through the whole chain, not from a cleverer model. When SCADA, forecasting and market data live in the same schema, you stop scoring the forecast against energy the plant was ordered not to produce. That single correction gave us more than adding more weather models ever did."

Key facts

Platform: DYNVOLT, solar SCADA, EMS and energy market intelligence software for utility-scale PV and storage, built on one data model from inverter string to market bid
Deployment: production plants in North Macedonia, 840+ PV strings monitored at 30-second intervals from a secured Linux edge server on site
Forecast accuracy: only 5.1% nMAE day-ahead, daytime hours, April to August 2026, normalized to AC capacity; methodology published
Event: EU PVSEC 2026 (43rd edition), WTC Rotterdam, Startup Pavilion and Startup Solar Launch Pad, 14 to 18 September

One edge server, one data model

DYNVOLT runs on a secured Linux-based edge server installed at the plant. Plant data lives on that server, buffered through outages and synchronized to the cloud for forecasting, dispatch and reporting. DYNVOLT never places bids or takes custody of trading revenue; schedules export for the owner or their balance responsible party to submit. DYNVOLT is now offered to asset owners, IPPs and traders across Europe.

About DYNVOLT

DYNVOLT is an AI-powered solar SCADA, energy management and market intelligence platform connecting utility-scale PV and storage plants to European energy markets on one edge-first stack and one data model. DYNVOLT publishes its forecast error, its worst day and what it has not built yet. Based in Kavadarci, North Macedonia. More at https://www.dynvolt.com

Media contact
Zoran Aleksikj, Co-Founder, DYNVOLT
zoran.a@dynvolt.com, +1 332-456-0669
Original release: https://www.dynvolt.com/blog/dynvolt-solar-scada-forecasting-platform-eu-pvsec-2026

Zoran Aleksikj
DYNVOLT
email us here
Visit us on social media:
LinkedIn