How to identify 'spike' automatically during data acquisition

Status
Not open for further replies.
T

tenderne

Member level 2
Joined
Jan 9, 2007
Messages
46
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,286
Activity points
1,621
Dear All,

It seems to be a 'weak' problem, but it really troubles me couple of days.

My problem is, I got some sets of real-time data from sensor, please see the attachment. Suppose I don't know the so-called 'threshold' value highlighted by red box, how to identify the 'spike' highlighted by green box?

Any hint would be much appreciated. thanks a lot in advance.

Best regards,
tenderne
 

Attachments

  • q1.png
    7.2 KB · Views: 164
  • q2.png
    6.3 KB · Views: 163

Take a moving averge (lowpass filter) followed by a peak detector to discriminate between the red and green box conditions
 

Dear klystron,

Thank you very much for your suggestion.

Actually I have tried moving average, after that a similar signal pattern is generated. however, with unknown threshold, I still cannot identify the 'spike' or 'peak'. please see attachment. in the figure, blue curve is original signal, red curve is the one after moving average.

Would you please elaborate more about the 'peak detector', assuming that we know nothing about the threshold? thanks a lot.

best regards,
tenderne
 

Attachments

  • q1_more.png
    17.7 KB · Views: 152

Calculate the slope (1 st derivitive) The peak occured where the slope change sign
 


This low pass filter is good for instantaneous data value (it filter noise). For running average use another low pass filter with a much lower pass band.
A similar algorithm is used for capacitive touch detection (search on Microchip web site, mTouch application notes describes this algorithm)
 

Status
Not open for further replies.

Similar threads

Menu
Log in
Register
Cookies are required to use this site. You must accept them to continue using the site. Learn more…