Back to overview​

Real-time data analysis in Analyzer4D

Our solutions are based on software and intelligent data analysis. Data collection only takes on its true meaning when it is analyzed. Our system is designed to display complex data transformations in an intuitive way and to carry out complex analyses during the measurement.

Our software is designed from the ground up to enable complex control, extensive data transformations and comprehensive analyses directly on the machine. At the same time, it is characterized by its flexibility and versatility. 
To this end, our software provides a multi-layered stack of tools that can be configured via our user interface:

  • Real-time logic and control unit (trigger list)
  • Hardware frequency analysis (FFT)
  • Frequency-specific filters​
  • Real-time compression​
  • Graphical programming of real-time analyses (operator network)
  • Python interface for data analysis

 With this comprehensive range of functions and tools, we enable highly efficient and technically sophisticated data evaluation and processing in real time.

Graphically programmable analysis pipeline Operatoren-Netz zur Schwellwertüberwachung Grafische Parametrierung Umfangreiche Dokumentation Container-Bausteine Echtzeit-Analyse während der Messung Simultaneous execution of calculations Python integration

Graphically programmable analysis pipeline

Every machine and every manufacturing process is unique. In order to evaluate the generated structure-borne sound signatures in a targeted manner, a flexible analysis option is of crucial importance. We have therefore integrated a modular real-time data analysis into our software. In our operator network, analysis modules (operators) can be combined into complex analyses using drag & drop.

The available modules range from basic mathematical operations, data and machine communication access to modules for pattern recognition.

Operator network for treshold value monitoring

The implementation of a basic threshold value evaluation of the measurement data requires only four analysis modules.

The use of the Data Stream Operator enables access to the spectral data within the analysis. The Energy Detector calculates the actual threshold value analysis.

The results obtained can be saved directly in the database using the "Save Results" operator.


Graphical parameterization

Almost all analysis modules provide several parameters that can be used to adjust the evaluation.

Double-click on the analysis module to open the interface for parameterization.


Comprehensive documentation

The analysis blocks represent complex calculations in the background. The operators have several inputs and outputs for linking with other modules. All more complex evaluations offer parameterization for fine adjustment.

In order to be able to use the operators in a targeted manner, the functionality and interaction with other modules must be known. Extensive help pages with explanations, calculation examples and illustrations with specific application examples are available in the software for the operators.

Container modules

So that you do not lose track of even complex analyses with many operators, analysis networks can be summarized in so-called "custom operators". These new, self-created operators can be saved as templates and reused multiple times and across projects.

This makes it easy to create analyses from more than 20 or even 50 operators.

Real-time analysis during the measurement

How does the evaluation of data work at all, even though the measurement has not yet been completed?

We have implemented a special evaluation logic in our operator network for this little trick. First of all, before execution, the sequence in which the blocks must be executed is calculated so that the data is available on time according to the links drawn.

The network then goes through several phases for the evaluation itself:

  1. Initialization phase at measurement start​
  2. Run phase during the measurement (repetitive)​
  3. Closing phase at the end of the measurement​

The main work takes place in the run phase, which is executed again for each new spectrum in the data stream. The operators process the new spectra and, if necessary, deliver results to subsequent blocks.

Thanks to the efficient implementation of the analysis modules, the entire analysis chain can be run through thousands of times per second.

Simultaneous execution of calculations

In order to be able to deliver the evaluation results as quickly as possible, we consistently use multithreading in our software, i.e. the genuinely simultaneous execution of calculations. The high-performance processors of Optimizer4D support the simultaneous execution of up to 16 parallel calculation threads.

In order to exploit this potential, we do almost everything at the same time and in parallel:

  • Data transfer from our measurement hardware to the software
  • The compression of data​
  • The use of frequency-selective filters​
  • Editing the control logic (trigger list)
  • The evaluation of analysis chains in the operator network

Even the evaluation of several data analyses can take place simultaneously. For example, we use parallel analyses for crack detection in bending straightening in order to detect component fractures even faster.

The software takes care of the parallelization for you!
 You only have to link the corresponding operators to exploit the full computing potential.

 

Python integration

We have integrated Python directly into our software. This means that the Python interpreter can execute Python code completely and without compromises.​

Using special interfaces (so-called bindings), Python can directly access the most important data structures of our software, for example:​

  • Data streams (buffer)​
  • Analysis results​
  • Database interface​
  • User interface (PenGUI)
In this way, completely new analysis methods can be used and methods can be developed outside the software, but then applied to the machine in real time.