Hi,
i am currently trying to approximate the distance of a sensor when being moved. I am recording all the raw sensor data for later analysis with python.
Those values are 16bit for XYZ-axis as well as an timestamp.
I am working with a discrimination window to neutralize all values which are below a calibrated threshold-value. Also I am using a butterworth lowpass with a 6Hz cutoff frequency.
When integrating the acceleration data twice my distance value is very far off.
I read a lot of the filtering already and that its very application-oriented but I am not really sure where I should start from right now.
Thanks for your help already.
It seems that you introduce an offset in the velocity which is integrated further.
Which is the movement you want to estimate ?
Which is the displacement you expect ?
Thanks for your answer,
this was just the sensor moved forward and backwards along the X-Axis. I ended the movement on the start position so the distance should return to near zero.
I also tried a homogeneous movement along the X-Axis but the distance also drifted far off the expected value.
Then you have a problem of offset due to filter and other unknown effect.
Try other filter strategies or a simple integration to see the effect and compare. I would like to see results for further analysis.
Thanks for your answer Nick,
I am still stuck with the selection of the filter. I also tried a Bandpass with a lowcutoff=0,244 and highcutoff=10Hz.
I am not quite sure how i should choose the filter parameters. Should I use a predefined distance where I move the sensor along and then tune the filter settings until they are as expected?
I have to say Im pretty new to this topic.
You get still an offset in velocity which leads to a slope in displacement after integration but it looks a lot better. I dare suggest an approach: if you have a wave generator give at your filter input a saw tooth voltage with your accelerometer amplitude + frequency and tweak the low and high pass thresholds till you get a correct output. Then if you recorded the true accelerometer output connect it to the filter and compare results. It will be a cut and try approach but it could help.
Viel Erfolg
Please visit this link below:
https://riunet.upv.es/bitstream/handle/10251/78186/PUCHALT%20-%20Evaluaci%C3%B3n%20de%20la%20medici%C3%B3n%20de%20distancia%20mediante%20un%20aceler%C3%B3metro%20digital%20de%203%20ejes.pdf?sequence=1
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Software