Best way to determine what is going on is to get a console connection when the problem is occuring and capturing:
show proc cpu >> what is the cpu util?
show spanning summary >> see if the STP root is what it should be.
show log >> look for any clues that will explain the cpu or STP.
show spanning-tree vlan x >> for each vlan.
if possible, disconnect the links one at a time, if you suspect a loop, start with uplinks or link connecting to another switch. Once connectivity is restore as soon as a particular link is severed, note what is connected on that link and reconnect the rest of the link one at a time to verify that the issue is gone. The next step would be to investigate why that particular switch is causing a loop.
