Dev: Negative values in power consumption

[stevebuk1]

The odd negative power consumption value appears after subtracting EV consumption from total consumption. Whilst this should have no effect, negative values are best suppressed to zero.

[stevebuk1]

Debug Logging shows the effect persists for a few hours but is short lived. Both Solis and Zappi sensors report a large power consumption for the most recent half hour. The Solis sensor reverts to a lower value a few optimiser iterations later, but the Zappi large value persists for a few hours and then reverts.

Rather than suppressing, a targeted logging has been added to Beta-4 that is triggered by any negative power values to get to root cause.

It is suspected that because this is occurring at the end of a charging session, the Zappi kWh sensor (1dec pl) is no longer incrementing meaning an absence of further entries. The existing consumption code probably performing an interpolation rather than assuming a zero increase. Once sufficient time has passed for the Zappi sensor to generate an entry, the problem resolves itself.

pv_opt.log pv_opt1.log

[stevebuk1]

Targetted logging suggests pandas .interpolate will not perform an extrapolate function and as such the root cause is that the EV from total subtraction is done on a power (W)) basis as opposed to an energy (kwh) basis. The subtraction really needs to be performed before conversion from kWh into W, which is an extensive fix. For a workaround any negative values will be zeroed and as the issue is always constrained to a single 1/2 hour slot the error introduced is unlikely to have any significant effect.