How to analyze 'non-linear ideal function'?

[Reliability Technology Enthusiast]

Good evening,

I am new to this forum. My name is Sasun working with 2-wheeler
company in India.

Recently, I was fortunate to attend to Mr Shin Taguchi's program on
'Robust Optimisation'. Eversince I attended that program, I got
fascinated to this technique of robust optimisation.

Quickly, I started adopting the 'energy thinking' methods for many of
the problems we are into. Interestingly, some of the problems for
which we had collected lots of data but could not do much - as the
results were 'very confusing' got astonishingly clear with 'energy
thinking' and using 'dynamic SNR' approach. The applications are:
Fatigue Life prediction.

Encouraged by this, since last 1 week, we are trying to adopt the same
technique for NVH problem. But the response data is higly non-linear
and I am not clear as to how I calculate SNR for this data. Here, I
have a 'reference data' form a benchmark model as an ideal function
that I need to realize with the design modificaions.

The ideal function is Engine RPM vs noise level in dB. I have 4
factors with 3 levels identified as control factors. And noise facotrs
identified are: 'age of the vehicle' - old and new and 'different
riding styles' - agressive, normal

I would be greatful if the 'steps involved' in calculating the SNR for
such non-linear situation is provided.

Thanks in advance

Sasun
--------