Hammerstein-Wiener system-nlhw

Hei,

I have the input and output signals. My point is that I have many experiment, for example ten experiment. Using one exerperiment data, we can having one nlhw model. Thereofe, we can have ten different models.

So is there any way that we can train this nlhw model by using not only one experiment signals but using all the expriment signals. One experiment we have (x_1(t), y_1(t)) t=1,...5000. With 10 experiments, we have (x_1(t), y_1(t)) t=1,...5000, (x_2(t), y_2(t)), t=1,...5000, .....(x_10(t), y_10(t)) t=1,...5000.

x_1, x_2, ....x_10 are the observation for the same input variable and y_1, ...y10 are the observation for the same output.

How can we train the nlhw model of using those ten experiments.

Any idea about this.

Thanks.