Neural Network Time Series, Multiple Inputs, Multiple Sequence input format

I cannot figure out the proper format to pass to a narx net for a dataset that I have. Please excuse my choice of variables but I have "n" inputs and "m" datasets with "i" samples in each set

I start with a numeric input matrix of size (i, n, m) and a target matrix of size (i,m).

I'm sure there's a simple answer out there but I would really appreciate if anyone could help me out because I am banging my head against the wall with this one!!!

I know I need to convert the num's to cells and do something along the lines of what they've done here: https://www.mathworks.com/help/nnet/ug/multiple-se...

But I can't figure out how to make that work with multiple inputs.

Thanks!!!

I also posted this in eng-tips so I'll cross-post the answer to both forums.

[Kamuran Turksoy]

Have you found an answer to your question?

[Matheus Torquato]

I also have a similar question. I have 200 cycles of data from 100 engines (with 24 parameters each).

I have organized my input matrix like this:

{Engine1_Time001}    {Engine2_Time001}            {Engine200_Time001}
{Engine1_Time002}    {Engine2_Time002}   .  .  .  {Engine200_Time002}
{Engine1_Time003}    {Engine2_Time003}            {Engine200_Time003}
 .                    .                  .  
 .                    .                     .
 .                    .                        .
{Engine1_Time200}    {Engine2_Time200}            {Engine200_Time100}

Where {EngineX_TimeY} is a 1x24 Cell containing 24 parameters of the Engine X at cycle Y.

The Neural Network Time Series App doesn't seem to be able to handle that structure since the options available are only Cell Column, Matrix Column and Matrix Row.

I would like to know how can I train this network (NARX, preferably)

[christttttttophe]

Did you ever figure this out?

Seems crazy that multiple sequences of data occurring at the same times cannot be analyzed with neural networks in Matlab. I'm beginning to find Matlab quite frustrating.

[Marvin Eckert]

I have the same question, all examples are with one time serious. How do I train on many different time seriers, like from different machines, creating on model.