Aiolos Forecast Studio 10.2
New module in Aiolos Forecast Studio to enhance intraday trading and integrate evaluation in the daily forecast work in Aiolos
- Integrated Evaluation – Learn from previous made forecasts displayed in Home together with your new forecast
- Enhanced intraday trading – Compare new forecast with previous, display differences in Home, Evaluate latest weather forecast with the previous
- Saved forecast corrections – see your colleagues forecasts inlcuding corrections
- Display regulation prices – see relevant market situation together with your forecast in Home
- HOME Live is a paid feature. Please contact email@example.com for more information
Availability Schedule improvements
Simplifications for running scheduler in Home when you have many series and the availability report sometimes comes in late for the user to handle.
- Filter template in Scheduler to choose a template of series to show. Create your own scheduler templates with series that you often work with
- Right click on series in series tree to open scheduler only for that series
- Improved performance for saving multiple changes in schedule at the same time
Improvements for DayType calendar
- Created import and export tool in Aio FileStudio to transfer daytype calendars between configurations
- Functionality in Aio FileStudio to dump a Legacy calendar into a text file
- Locked basic daytype rules to prevent users to make unwanted changes in the daytype behaviour
- In daytype calendar properties view you can now ignore collisions that are solved with priority in daytype
The new secure file transfer uses a restful API to send and receive data files between the customer´s network and the Vitec server. The data transfer is used for our hosted enviroment to send and receive data and forecasts in a secure way. The datatransfer can be used for up to 4 GB files. Connection is independent of platform on the client side which means that it can be made both from a Linux or a Windows server.
The Secure FileTransfer is a secure way of transferring data and all that is needed on the customer side is a small client or if you choose to run an open powershell script to send and receive data. The source code for the client program is open for review by the user.
The client can be fetched at: http://energyappdeploy.vitec.net/
- Secure FileTransfer is a paid feature. Please contact firstname.lastname@example.org for more information
Market data improvements
- Portfolio effect is now accounted for in FollowUp Market data for aggregated series
- Sum between different price areas is now possible for aggregated series
- Regulation price information integrated with the new module Home Live
- Import of Dominating Direction to get the valid regulating price for different hours. Used for calculating imbalance cost for hours with both up and down regulation
Map in Config
Changed the integrated web browser to chrome because .Net stopped supporting Internet Explorer
Aiolos Quality Service improvements
Improved filter for the sort of error to log helps to only show relevant errors in Status and in e-mail. The flag “disregard error” lets you choose irrelevant errors to ignore
- “No events found” warning can be ignored so that the import does not warn for missing status items in case this information is not relevant for the user
- “No Column with header” warning in the import can be ignored if this information is not relevant to the user
Improved sorting in series tree
- Sorting of series on volume. It is now possible to sort the largest series on the first positions in the series tree
- Improved performance when sorting the series tree
- Sort without restart of Aiolos, toggle between different sort orders in Home
- Base model always visible to the left in tree and grid
- Performance slightly improved for reading AutoTune files
- Number of results that can be loaded in AutoTune is maximum set to 100
- Fixed comparison problems between model families
Micellanious functionality updates
- Download Saved forecasts
- In system settings choose how many days to save forecasts
- Improved messages from “download saved forecasts”-tool in Home
- In zip backup configuration choose if saved forecasts should be included in the zip-file.
- Increased warning level in data status check for errors in weather observations
- Simulate Adjusted weather with smoothening in View
- Changed design in Edit Fractions in Config, the functionality for create and edit are separated
- Added “Hard reset docking windows” in File where the stats-files are completely removed instead of only restored
Simplifications in Config
- Autosearch for weather when creating new forecast series is improved with settings for simple startup weights of weather stations
- Config cleanup improved design
- New filter functionality for the grid. Filter out Template in config – Grid controls
- Improved filter out decendants and filter out non-children
Master Data Sync improvements
Automatic master data syncronisation and/or import of forecast series from another database, excel or other source. Functionality improvements made in this release:
- Works for forecast series names only containing numbers
- The user can choose to synchronize only on forecast series and leave aggregated series as they are in Aiolos, which allows import from CRM system
- Master Data Sync can use static settings in the sync protocol using default value e.g. an import protocol
- History start up setting added in user settings
- HOME Live settings found in system settings
- Write follow up log in system settings has a time span until automatic reset. This to prevent large log files to be stored on the server if not used
- Improved design with groups in user settings
- Improved performance when forecasting 10.000+ series
Improvements in Follow Up
- Recreate forecasts in follow up works with models that use real time corrections
- Exports from follow up to csv/xlsx or other format
- 328 performed updates and fixes from the last version of Aiolos Forecast Studio.
A clear aim has been set in the development team to improve forecast accuracy. Improvements has been made for models forecasting hydro, wind power, solar power and load forecast models. Functionality for advanced realtime correction has been implemented within the Model Studio suite that can be used for any model type or even as stand alone short-term modeling methods. In addition, functionality for automatic backtesting and optimization for the choice of weather point has been implemented in Model Studio.
Aiolos Model Studio
Optimize weather point
- Find best weather point for forecast series. Optimization tools for automatic analysis of weather stations in the vincinity of the forecast series to tune the best choice of weather.
- Autosearch improved in Weather Client to fetch and create a number of closest weather stations.
New Script shell models
Shell models that selects data passed on to the script. These shell models enables low level control and end user modeling in Model Studio.
- Seasonal model – model that uses the seasonal data the recent years as training data
- Industry model – model that uses the x most recent days of the forecasted daytypes as training data
- Realtime model – model that uses the recent period together with the forecasting period as training data
New script models
- Live Correct model – Auto correcting model based on the most recent forecast errors. Gets the errors for the latest days and self-corrects the forecast as adjusted forecast in Aiolos.
- Interactive Industry model – model where interactice forecasting/modeling is enabled by Aiolos Forecast Studio by a graphical user interface plug-in
Advanced Realtime Corrections
When using Model studio script models in Aiolos – Realtime corrections can be based on any of the methods below. The Model Studio code for real-time corrections is transparent and editable by the user.
- Autoregressive (AR)
- Autoregressive with exogenous variables (ARX)
- Simple (or Single) exponential smoothing
- Double exponential smoothing (Holt)
- Linear double exponential smoothing (Brown)
- Exponential smoothing with a damped additive trend (Gardner-McKenzie)
- Exponential smoothing with a damped multiplicative trend (Taylor)
- Triple exponential smoothing (Holt-Winter)
- Moving average
- Moving median
- Customized methods
The indata for the real-time correction has been improved. In the basic realtime correction only the latest historic value is used for correction, in advanced realtime correction the training period for correction is a user set time-span. The period of correction applied on the forecast can be set to longer than 24 hours, which is the maximum value for the basic realtime correction.
Normally the realtime correction is a last step on a larger model but the advanced realtime correction methods alone can also be used as Short-term intraday forecast methods.