Hi,
I am currently exploring the design of APRBS (amplitude modulated pseudo random signal) sequence. At present the one I have developed was based on literal word of signal, i.e. using the PRBS and performing amplitude modulation in order to achieve an APRBS sequence.
However, I am hoping to find a substantial methodology which allow to manipulate the amplitude and frequency at different regions.
The purpose for signal design is to create training data for neuro-fuzzy model which is able to excite system of interest at all realistically possible frequencies.
As I have failed to find appropriate and relevant resources, I was hoping if anyone could point me in the right direction.
I appreciate any comments and help.
Kind Regards
Gaurav
For an interesting theoretical view of optimal test signals you may be interested in this paper:
Fang, K.; Shenton, A.T., "Constrained Optimal Test Signal Design for Improved Prediction Error," Automation Science and Engineering, IEEE Transactions on , vol.11, no.4, pp.1191,1202, Oct. 2014
http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6558486
On a practical level, I have had some success tuning APRBS signals by superposition of several PRBS signals with different amplitude and sampling rate characteristics, but it isn't a scientific approach.
Hi Christian,
Thank you very much for taking time to reply.
Indeed, it is an interesting paper and presents a wide range of information on the topic. I am grateful for the link.
I am neither in position nor stature to comment on scientific approach of your method but I would like to think that it could provide a pragmatic solution .If you would allow me, I would like to try and implement your approach for tuning APRBS signals.
Gurav,
Please, go ahead and try it.
One other thing you might like to try is to numerically integrate the output from a PRBS generator. This will result in a kind of random walk signal which will have a lower frequency component with varying amplitude. You can then superimpose one or more PRBS signals to it, which will add a 'high' frequency component. The resulting signal will then have a non-zero mean.
Christian
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Programming Languages
- Python