That's a headscratcher all right.
For 2 years operating, about 1 month to 2 months operating, motor is trip by single phase overload (always random 2 phase ~ 450A, 1 phase ~ 1000A).
Are you saying that the motor has operated for 2 years normally and then the problem only showed up in the last 1-2 months?
I can appreciate the logic of why a ground fault would be suspected, but…. 500A+ ground fault jumping among phases and doesn't leave any evidence (burn marks) or trip any protective relaying other than overload? That seems unlikely, although I don't have another ready explanation.
If the machine were delta-wound and there were for some reason open circuits in one leg, then we see one leg carries the vector sum of the other two leg currents. But why would an open circuit jump among phases?
Some unusual things might happen during transient. Is this trip occuring during a start? During some other transient? Or during steady state?
How are these reported currents measured / determined? Read from a digital relay? Measured from a clamp-on?
What is the grounding configuration of the supply to this bus? Solid grounded? Resistance Grounded? Ungrounded? Any ground protection for the bus?
Some troubleshooting to consider:
Check for three-phase current balance during steady state operation.
Get oscilloscope traces of currents if transient or oscillation is suspsected.
Megger and bridge stator. Include as much of the supply circuit as possible.
Megger and bridge rotor circuit.
Check resistor switches.
Investigate the ct secondary circuit. Typically has only one ground at center of a wye connection on the ct secondary. Lift the ground and megger remainder of circuit to ground at 250vdc (well at least we do that with our electromagnetic relays… you might want to take any electronic relays out of the circuit before that test).