Assuming that the drive is being told to start/stop over Ethernet, I'd bet that the issue is how you are issuing the start/stop commands.
A common method people use is to use the "not start" to issue the stop command. This will occasionally causes the issue you are experiencing. The reason being, Ethernet communications is asynchronous to the PLC scan, so every once in a while, if the comms updates at the point between your start command going away and the stop command being issued, the start and stop will be true at the same time, causing the drive no start even though you see the start command on at the same time.
A good practice is to only issue the start and stop commands as long as they need to be to get the drive started and stopped. An easy was is to use the drive active bit. The start command doesn't need to be on continuously. So putting the "drive not active" in front of your start command and "drive active" in front of your stop command will ensure that the commands can't be on at the same time.