What's new in version 4.0 of ANDATA software tools?
by Andreas Kuhn
Version 4.0 of ANDATA's software comes with some fundamental changes and enhancements, which are shortly described here in the following.
Attributes in Stipulator
The complete internal data model and data processing has been refactored in Stipulator. The most significant change is the introduction of the new "Attribute Processing" concept. The former "Numbers" and "Categories" are now combined to uniform "Attributs", which can be assigned with individual properties. In addition a new "Text" attribute for the specification of a "Requirement" has been defined. In the same way "Links" can be assigned, which allow the connection of requirements to an external document. For example simulation results of videos from test can be opened directly from Stipulator that way. Beside the administrational advandages attributes can also be used for signal processing. That way attributes with variable, scalar values can be propageted as input into the signal processing, which can then be used within arbitrary and individual signal processings. In a similar way attributes can be calculated from signal processing and assigned to the attribute values in Stipulator.
All tools and all their internal dialogs have been refactored with respect to usability. Also the handling of all the processes have been unified and harmonized as far as possible.
All internal calculations in all the tools have been optimized with repsect to speed as far as possible. For that the Parallel Computing Toolbox of MATLAB has been utilized, everywhere this makes sense. This covers signal processing and data extraction in Stipulator as well as the training in Brainer. Also the parallelisation in SceneInspector has been improved. Simulations can now be easily distributed to local workstation clusters.
But not only calculations have been parallelized. Also the internal data structure and the memory management has been refactored. Especially in the wake of huge data sets (refering the Big Data trend), the data matrices can now be treated in seperate data blocks and files, being processed seperately. This approach makes the adaptive training approach much easier, which already has been implemented in Brainer and will be released for the different models within the following releases.
Version 4.0 comes with the completely new implemented software tool Expectator, which is a programm for the design, administration and execution of automized anomalie and incident detection for arbitrary (signal) data.
Generalization of Signal Processing
The signal processing in Stipulator had been strictly separated into the three sections "raw", 'basic", "processing" so far. This strict and fixed separation has been released now. In the upcoming version of Stipulator "Signal Processing Sections" will be defineable and combineable arbitrarily. This has the advantage that several signal processing steps, which are used more often, can be collected into an own "section" and management individually. In future releases these sections can be stored and used in different signal processing sequences. That way complicated custom signal processing functions can also be managed by the signal processing definition dialog.
Non-Equidistant Time Series
The Signal Structure Toolbox is now able to read and treat signals with non eqidistant time channels. That way also event based data can be treated much more efficiently. The data can then also be stored in the Signal Structure format.
Simulation Engines in SceneInspector
The simulation engines in SceneInspector have been completely refactored. Expecially this has been done for the PC-Crash engine as well as for the VISSIM engine. The PC-Crash engine now uses the new features of PC-Crash version 10 and is capable of probabilistic in-the-loop simulation.
Further details can be found in the user manuals as well as in upcoming blogs at this location.