ROVI-org / auto-soh / 10836905619

95%

push

github

Degradation model (#85)

* Creating base DegradationModel

Base class to establish the minimum requirements of a degradation model. Children
of this class maybe need additional inputs, and may need to store previous states
as well.

* Creating dummy degradation model

Degradation model that does not degrade the A-SOH.

* Allowing transient history to be passed to degradation model

* Adding test for dummy degradation

* minor modifications to base degradation

* Allowing outputs to be passed to degradation as well

* Fixing model tests

We now require CellModels to be passed to the simulator with the 'cell_model'
keyword argument (or as a positional argument too), so some tests had to be
fixed.

* Adding degradation model to simulator

* Fixing estimator tests

The Simulator no longer takes a 'model' keyword argument, instead it takes a
'cell_model' one.

* Fixing notebooks

Since simulator now requires 'cell_model' instead of 'model', notebooks had to
be fixed.

* Adding degradation to ModelWrappers

Both dual and joint model wrappers now have degradation.

* Fixing bug on interpolator

Found a bug while testing the degradation model with the joint filter. In short,
calls to the ECM.update_transient_state where altering the value of the base
values of the H0 attribute of the A-SOH. After some investigation, I found out
it was happening in the special case where interpolation was not necessary: we
needed to copy the base values, otherwise, we would be changing the base_values
in the H0. That seems to have fixed the issue.

* rerunning joint demo after degradation changes

* Making sure degradation works with joint estimation

* fixing typo

* Fixing bug in interpolation

The SOCInterpolation would add a dimension when we didn't want it to. In the case
where the SOC was passed as a 2D array of shape (1,1), and the health variable
batch size was N, the returned array would be of shape (N,1,1), w... (continued)

29 of 30 new or added lines in 5 files covered. (96.67%)

1282 of 1359 relevant lines covered (94.33%)

0.94 hits per line