I want to calibrate accelerometer data with force platform. I try more than filter but I want to know what is the best one for keeping data true.
This depends on the activity, the sampling frequency and where you are placing the accelerometers. Force plate filtering is typically at 25Hz for walking with a sampling frequency of 200Hz. What are you wanting to look at?
i do comparing between bertec force platform the sampling frequency 1000 hz .it is using anti-aliasing filter 500 hz. and i have beanair wireless 3 axis accelerometer 1000 hz it is using antialiasing filter 10 hz . i put the accelerometer in the center of body . i try to do vertical jump . Curves in of both device are closely same but i have more noise in accelerometer data and values big than force plat form values ( i get force by inverse dynamic F = m.a ) i used smoothing filter by averaging 300 point values now are so close to compare the peak of ground reaction force of both device .but this way unstable some times 300 point it can be increase to decrease in the values I used the manual of both device to know the specialists of both. I need the right filter to process my data
Are you trying to replace the force plates with the accelerometers? Over filtering will have the effect of reducing the peaks. I would suggest a butterworth filter for both data sets. Are you interested in impact? if so a low pass filter set at 100 Hz may be a good place to start. If you are interested more in the CoM movement then I would go lower. I suggest you look at some papers to see what they do for vertical jump as this is not something we do very often in my lab.
Can you look to this file . Accelerometer vs force platform .in three trial of vertical jump
What filter have you applied to these?
What is the question you are trying to answer with the study?
The acceleration data certainly needs a different filter than the one you have used here.The force data looks okay, but not sure how you have filtered. Is this with a 300 point moving average? I would not recommend this where you have got impact forces which may be of interest.
the question of this study ? can i use accelerometer to assess peak of power without force platform . because the accelerometer has a good advantage ( wireless 500 m - small - cheaper than force platform - record ground reaction force for every movement -makes athlete feel free when he is jumping or running ).this is the first question of my study - second question what is the different between three level of athletes in exclusive power by using accelerometer.
this is the Raw accelerometer data .i don't use a filter with it .but force platform data is automatically filtered with the software . i will attach you comparing image 1- raw accelerometer data 2- accelerometer moving average data 300pt- force platform data
Mansour,
The calculation of peak power during a jump from accelerometer data could be very sensitive to the inertial characteristics of the task. Mass will have to enter into the calculation somewhere and, in this case, the mass that the accelerometer is tracking will actually vary over time. This could potentially confound measurements of peak power, especially when comparing across different subjects. Average power over multiple jumps could be much more robust, such as described here:
http://www.ncbi.nlm.nih.gov/pubmed/6681758
I agree with Dr Richards that you will probably want to use a filter that preserves the high-frequency content of the signal. If you have access to Matlab, there are some very powerful filtering tools that can be useful for your purposes. A recursive Butterworth filter is often used, but I have also had success using wavelet filtering, which can remove very tight bandwidths of noise without affecting the signal too much.
Hello sean,
I have tried more than filter but all values it seam bigger than force platform values .could you do me a favor please . can you possess this data with right filter because i am not good in this field . i will send you the Raw accelerometer data with (g unite measurement ) 1 g= 9.8 m/s^2 sampling rate 1000 Hz its using anti aliasing filter analog to digital 10 hz automatically form software . athlete mass 76 kg and force platform data for this athlete 1000 Hz but its using anti aliasing filter analog to digital 500 Hz automatically form software . i am really confused with these filters. i am very appropriate you if you do this possessing
The accelerometer the CoM times body mass, F=ma, will not give you the same as the force plates as it will not be considering the inertial properties on the lower limb which will be experiencing larger inertial forces, Force on segment = mass of segment x acceleration of segment. This would need to considered for the foot, shank and thigh. Not to mention movement artefacts and the factors Sean correctly alks about.
The reason I asked why are you doing this is key. Power is useful but other measures from the accelerometer could be equally as useful e.g. peak acceleration, work done (equivalent to average power Sean mentioned) and height jump can all be estimated from accelerometer data.
You are right Dr Richards.but can i predict the peak of power from peak of acceleration. I mean the peak of acceleration with the time like indicators for exclusive power as equation P= F.v
i use acceleration here as indicator for force and time is indicator for velocity
the athlete who has a big peak of acceleration in little time he has high exclusive power
PubMed yields 8 papers on "vertical jump accelerometer" a couple of these look relevant. You should also search sports discus.
Very happy to try and help. Do let me know how you get on. Best wishes.
I think Residual analysis Winter (1990) until now is a nice way to identify the optimal cut off frequency
- Badges
- Science topic
- Similar topics
- Mathematics
- Algebra
- Matrix