Hi,
I have a time series problem. I want to find a model for a time series which has multiple outputs. I have tried using ARX as so...
SAD1=ModelPsad(1:500,1:10); % this is my time series data h1=iddata(SAD1,[]); % make it what I think is the correct input format for iddata model= arx(h1, 'na', 2*ones(10)); % create model.
I use predict and compare functions and the model adequacy seems to be good, with good fit for the models.
Next I want to simulate and generate a new time series with the generated model -- and check whether the synthesised time series is similar to the original. I originally thought I would use simulate as so...
input=randn(size(SAD1)); y = sim(model, input);
However the output y is really quite rubbish, as doesn't look like the original . So either this means the generated model is not adequate, or
I am possibly simulating not in the correct way. Is this the correct way to synthesis a new sequence with matlab system id toolbox????
I have tried using Arfit with my data, and simulating with this I get a
good simulated output but the adequacy tests say the models are inadequate.
Any help on this problem and any suggestions would be greatly appreciated.
Nanette