Advanced Data Analysis Pattern Recognition & Neural Networks Software for Acoustic Emission Applications. Topic: Waveforms in Noesis

Transcription

1 Advanced Data Analysis Pattern Recognition & Neural Networks Software for Acoustic Emission Applications Topic: Waveforms in Noesis 1

2 Noesis Waveforms Capabilities Noesis main features relating to Waveforms: 1. Import of multiple waveforms from several format 2. Export of waveforms to several formats 3. Plotting of waveforms using different representations (time and frequency domain) 4. Advanced waveform processing (DSP) 5. Waveform statistical processing 6. Feature extraction from recorded waveforms 7. Waveforms and Features correlation (also Waveforms and Pattern Recognition Classes Correlation) 8. Live processing and Feature Extraction 2

3 Import/Export Waveforms Waveforms can be loaded into Noesis by import: AE Data (DTA) from Spartran, Mistras, DiSP, Sensor Highway, PocketAE PAC s Systems AE Wave Streaming Data (WFS) from PCI-2 systems. AE Wave Data (TDA): from TRA, MISTRAS and DiSP systems. ASCII Waveform (TXT) from tab delimited ASCII waveform files. Waveforms can be exported from Noesis: To their original format after processing To ASCII Waveform (TXT) Also WFS can be exported to DTA after splitting to shorter length 3

4 Plotting Waveforms (1) All waveform plots can be fully customized (from plot type to axes scaling and feature to font type and size) using the plot properties dialog, with just a simple mouse right-click. A large variety of plots are available. Pages set-up resulting in multiple SCREENS with any combination of graphs. 4

5 Plotting Waveforms (2) Time Domain Waveform plot type available: Time Domain Time - Wave Time Rectified Wave Time Wave RMS Time Auto Correlation Time - Cross Correlation 5

6 Frequency Waveform plot type available: Frequency Domain FFT - Real FFT - Imaginary FFT - Magnitude FFT - Phase FFT Power Spectrum Plotting Waveforms (3) 6

7 Plotting Waveforms (4) Advanced Frequency Waveform plot type available: Frequency Domain (more) Short Time FFT Short Time FFT Spectrogram Discrete Wavelet Transformation Continuous Wavelet Transformation Power Cepstrum 7

8 Plotting Waveforms (5) Statistical Processing and plotting of Waveforms Waveform Data Raw Data Moving Average Standard Deviation Skewness Kurtosis 8

9 Plotting Waveforms (6) Segment Waveform - Frequency Domain Views. The user can define a segment of the waveform and get the Frequency Domain plots for this segment. 9

10 Plotting Waveforms (7) Signal Processing (DSP): The user can apply frequency filtering and windowing to any wave view. The parameters are: Windowing type, Filter-Design, Filter-Type (Band-Pass etc), Filter-Order and Low and High Frequency. 10

11 Navigation between Waveforms and Plots Dedicated toolbar for: 1. Change immediately plot type 2. Navigation between waveforms (First, Next, Previous, Last) 3. Select navigation mode: Global Navigation: Synchronize all waveform views of a data set so that they display the same waveform Auto Select First: Navigating through the waveforms will cause the record of the first waveform to be selected. Mouse Select: Once activated the user can double click on any waveform and the corresponding record will be selected. 11

12 Feature Extraction (1) Feature extraction for each waveform with user defined settings. Note: Multiple hit extraction - Noesis will actually perform a Post Test Acquisition and will extract hits using the new settings. 12

13 Feature Extraction (2) Different settings could be applied to each waveform The user could define the feature he want to extract 13

14 Feature Extraction (3) Option to keep original waveform Option to split waveform to any desired length Application - Noesis Feature Extraction of a very large WFS. 14

15 Live-SPR. Real-time data classification and processing. (1) WFS files: Noesis can load WFS files as they are acquired. Depending on classifier Noesis will extract features and break-down a single WFS wave to multi-hits and classify the data. 15

16 Live-SPR. Real-time data classification and processing. (2) In parallel with live classification Noesis can compute a variety of Periodic Statistics that follow cluster evolution based on calculated parameters realtime. This feature is also available during post-processing. 16

17 Live-SPR. Real-time data classification and processing. (3) Data scatter plot with the data classified immediately with acquisition. Class velocity graphs, showing how fast the center of one class approaches or moves away from the center of the reference class or point. Cluster distance, showing the evolution of cluster center distances in time. Cluster size, showing the evolution of the cluster data in time. 17