System Identification impulseset function: how do I avoid using deprecated features?

To use impulseest to model a system, I run the following:

sstrain=iddata(target_train,inputs_train);
Imp = impulseest(sstrain,[-5 40],'PW',10,'noncausal');

I get this message:

"Warning: Use of a time vector as an input to the "impulseest" command is deprecated and may not be supported in future. Type "help impulseest" for more information."

"help impulseest" in fact has no information on this. Using a number as the second argument gives a similar warning. When I run impulseest with [] as the second argument, the result has a poorer fit. BTW, apparently the 'noncausal' option also has been deprecated.

I don't want to lose functionality in the future. What is the correct way to proceed, but still get the same quality of results I am getting now?

Thanks,

Kevin

Answer 1

[Star Strider]

The iddata call contains all the time information the identification functions need in order to estimate the system, in the array lengths of the input and output data, and the sampling interval (that defaults to 1 if not specifically provided). I would only use the iddata output with whatever other arguments impulseest needs to estimate the system to the desired specifications. A time vector or other time information is not necessary.

[Kevin Johnson]

Thank you. Kevin

[Star Strider]

As always, my pleasure!