Data Analysis
Numerical Analysis
There is one data file for each test series (power on to power off sequence) Each recorded accelerometer data file containing the header and sensor data is recorded in binary format and can be read with a customized data analysis program such as MatLab. The program was written to analyze a brief time period (10 seconds) centered on a time associated with the test body motion for a sensor of interest.
By performing an FFT analysis in MatLab, the accelerometer values are converted from the time domain to the frequency domain yielding a vibration frequency response for the processed time period.
Although not performed in this testing, time alignments from a standardized calibrated source such as GPS or IRIG time standards could be incorporated into the data stream header to more closely coordinate the motion of the test body and the sensor data.
Test Body Motion Analysis
The movements of the mass body as it progressed through a motion test must be recorded by an external measurement system. Items such as speed, G force, altitude, depth, temperature, humidity, etc need to be tracked in time and compared with the accelerometer data.
10 Seconds of Mass Body Motion as Recorded From Test Vehicle Data Bus
The above chart shows the mass body movement in speed, G force, and altitude for a 10-second time period. The figure below shows the same 10-second time period recorded accelerometer data and the associated frequency response spectrum. The data indicates vibrations from 100 Hz up to approximately 1 kHz with the peak vibration at 700 Hz. The frequency spectrum ends at 5 kHz because the anti-aliasing filter cuts the response of at that point.
The analysis shown is for one axis on one 3-axis accelerometer. Similar plots are available from the other axis on the sensor as well as other sensors on the mass body. The number of sensors times and the number of time periods of interest generates a large amount of data. The analysis from one mass body test must be subdivided into time slices selected for both time position in the test and the duration of the data for the FFT analysis.
It was found that giving the designer the ability to select sensors and time values in a graphical user's interface simplified the task of data analysis. Working from plots of the mass body external speed and position test data the designer easily selected sensor and time periods for analysis. The MatlLab program and user's interface generated all the frequency response charts as indicated below on demand quickly and easily.
10 Seconds of Accelerometer Data Centered at 6 hours, 58 Minutes, 30 Seconds
Page: 1 | 2 | 3 | 4 | 5 | 6 | NEXT
|
