metaScript Master
Customised forecasting logic for load profiles that go beyond the standard
Industrial customers with shift-based operations, variable production schedules and bespoke load management strategies do not fit into standard models. metaScript Master combines machine learning with your expert knowledge: your own, programmable logic intervenes directly in the forecasting process, precisely where the model alone cannot proceed any further.
Four points of intervention in the forecasting process
Most forecasting systems are a ‘black box’: data goes in, a forecast comes out. metaScript Master combines the forecasting model with your own logic at four stages of the forecasting process:
Proprietary forecasting method
The script handles the forecasting process entirely for delivery points with no historical data or where there are breaks in the data series.
Forecast Script
Accesses the results of an existing model and adjusts them. The model provides the baseline forecast; the script adjusts for known deviations.
Predictor Script
Generates a custom time series as an additional explanatory variable for the forecasting model, for customers whose behaviour depends on non-standardised factors.
Strategy Script
Determines which stored forecast model is used for a given point in time – e.g. summer vs. winter model, shift A vs. shift B.
Practical examples
- Delivery point with no measurement history: Synthetic standard profile as a starting point, gradually replaced by actual measured values
- Short-term changes in behaviour: A major customer changes their operations – the Forecast Script corrects the existing model without the need for retraining
- Structural breaks: New plant or capacity expansion – the Predictor Script generates a synthetic time series that reflects the new consumption pattern
- Calendar patterns: The Strategy Script automatically selects the correct model for summer, winter, public holidays and long weekends
Advanced forecasting functionality
- Limiting forecast results: By setting upper and lower limits in advance, forecast errors caused by measurement errors or transmission errors can be ruled out
- Value updating: Transferring and updating a measured hourly value for any number of hours or days into the future
- Application of defined profiles: A recognised historical load profile can be rolled out over any number of days into the future
- Extension of load structures: A load profile for a specific period can be extended for any number of days into the future
When you need the metaScript Master
- Industrial customers with variable production schedules or shift work
- Large-scale consumers with their own generation capacity and complex self-consumption patterns
- Portfolios with structural breaks or missing consumption data
- Consumers whose behaviour depends on factors not covered by any standard model
Forecasting logic tailored to your requirements
See for yourself – in a personalised demo, we’ll show you mP Xnergy in action