search

floracharbo / MARL_local_electricity

Multi-agent reinforcement learning for privacy-preserving, scalable residential energy flexibility coordination
GNU Affero General Public License v3.0
22 stars 5 forks source link

fix names recorded from network when manage_voltage #33

Closed floracharbo closed 1 year ago

floracharbo commented 1 year ago

These bugs for when manage_voltage = True could not be solved in PR #13 as there were optimisation errors that prevented me from running this setting at all.

julie-vienne commented 1 year ago

@floracharbo thanks for correcting this.

julie-vienne commented 1 year ago

Right now, we're replacing the voltage values if abs_voltage_diff > 1e-4 or rel_voltage_diff > 0.1). I think this might be a mistake. We should be making sure that the relative voltage difference is smaller than 1% (0.01) and not 10%, correct? Also, I would like to correct the count of voltage errors like this to count of absolute and relative differences:

        if replace_with_pp_simulation:
            self.n_voltage_error += 1

We use n_voltage_error to print a warning message but it should actually be the same as count_correction_opti_with_pp that we print in record.

julie-vienne commented 1 year ago

Also, I would like to correct the count of voltage errors like this to count of absolute and relative differences:

        if replace_with_pp_simulation:
            self.n_voltage_error += 1

We use n_voltage_error to print a warning message but it should actually be the same as count_correction_opti_with_pp that we print in record.

Nevermind this part, we're counting two different things with two different variables: the number of time we're replacing the values and the number of time the impact on the total costs in non-negligible. All good like this :) Except for the part where we're comparing the 10%

floracharbo commented 1 year ago

Right now, we're replacing the voltage values if abs_voltage_diff > 1e-4 or rel_voltage_diff > 0.1). I think this might be a mistake. We should be making sure that the relative voltage difference is smaller than 1% (0.01) and not 10%, correct?

these should be set as input parameters in grd what tolerance you want to set.