What commands do I use for using Time Series NARX Neural Networks for any random input?

I am new to neural networks. Somehow I have been able to make a neural network model based on NARX and it is working good for the predefined data that I used for its training. The input data is of 2 elements have 21600 samples and output data is one element of 21600 samples. Now I need to simulate this network for any random input. I don't know the commands for it. Any help will be appreciated.

Answer 1

[Greg Heath]

From the documentation

y = net(X, Xi, Ai).

Therefore, you need to also know the corresponding initial input and output feedback values in the delay buffers.

Therefore, this input cannot be truly random.

You might have better luck with NARNET where you can better estimate new initial conditions.

This is similar to asking what is the output of a known nth order linear difference (or even differential) equation from an unknown input if you don't know the initial values of n-1 lower derivatives.

Hope this helps.

Thank you for formally accepting my answer

Greg