You can try using projection pursuit principle and the resulting projection based outlyingness function.

Here is a reference you can start with:

https://projecteuclid.org/download/pdf_1/euclid.aos/1176349519

I would suggest to use the first n partial correlation coefficients to derive the low order model. Look up in the literature for the equations to convert the AR coefficients into partial correlations and vice versa. Note that ometimes the AR coefficients are called linear prediction coefficients (LPC) and the partial correlation coefficients may be called reflection coefficients. You can convert the AR coefficients into partial correlation coefficients, then take the the first n partial correlation coefficients and convert them back into an AR coefficients or order n.

Thank all for your answers!

Raziel Haimi-Cohen Thanks, it is exactly what I did. The process is known as Yule Walker equation if I am right.