Is it true that power should be calculated with Vdc if the signal has
no negative part?
As in all cases, power is the average of instantaneous voltage multiplied by instantaneous current be it DC or AC or any waveform shape or amplitude.
This means that if your DC rectified voltage has ripple, the current into a resistive load will also have ripple and that means that the power value is modified by that ripple.
But, if you know the true RMS of the voltage (DC with ripple), you can calculate the power into a known resistance as \$V^2/R\$.
However, you cannot accurately calculate power into the resistor based on the DC content of the voltage waveform except when the ripple is small i.e. the error will be acceptable in most cases.
I think it should be calculated using Vdc because the signal is close
to a DC signal, but am I right?
If the ripple magnitude is small compared to the DC level then the error will be small.
Just in case there is any confusion as to what a 3 phase rectified output voltage looks like, here's a picture of a standard 3 phase half-wave rectified voltage waveform: -
Picture from here although I've reduced it in size to remove the erroneous depiction of the full-wave rectified waveform.