Short Term Load Forecasting
Short Term Load Forecasting (STLF) utilizes historical load and weather data to forecast the system load. It runs automatically at the beginning of each hour and prepares a 168–hour forecast. It can also be executed on demand to provide a 14–day forecast, beginning from the current day. Results obtained at each execution are saved in the real–time database for use by other application programs.
The load forecast model is customized for each application. This customization process is comprised of two steps: initialization and maintenance. The initialization of the model parameters requires a minimum of eight weeks of seed data:
- load data
- dry bulb temperature
- wind speed
Subsequent to building the initial model, STLF captures the new values from the real–time database and saves them for up to one year. The model’s internal parameters are automatically adjusted. These adjustments also take into account any changes, such as load growth, in system conditions.
In addition, time and date can be altered on an STLF display. The program can be executed on demand and forecast the load for periods other than the current day. Components that contribute to the forecast are deterministic, base load (including weather–sensitive load) and uncertainty.
Based on the load component to be forecast, STLF uses different techniques:
- An exponential smoothing regression model.
- An auto–regressive integrated moving average model.
The model parameters are estimated and adjusted on–line by using recursive least squares (RLS) techniques. Input of STLF is obtained from and the computed results are saved in historic files and the real–time database. Data can also be stored in an SQL–type database, such as Oracle.
STLF user interface
STLF is provided with a number of sample display pages, such as a control page, summary page, holiday (or exceptions) page, etc. These may be used as templates and can be modified by the user. Since all input and output are resident in the real–time database, however, users are free to develop displays for their own use.
Short Term Load Forecasting Report Window
The STLF report display presents tabular information, trends, and control windows. The data in the tabular window can be updated to account for any loss of data or data that may be erroneous due to bad-telemetered data from the field devices. In addition, the columns can be adjusted so similar data can be dragged from one location to the next for ease of presentation. Also, control tabs are available to allow the user to select viewable data for the video trend.
Curve Selection Window
Forecasted Weather Display Window