I had a similar issue, I don't know the reason why the link between the RUTX and the switch goes down but a workaround would be to shutdown the port the switch is connected to and re-enable it if a ping of one of the WRT3200ACM fails.
1) add a cron job to be run every minute:
*/1 * * * * /bin/ping -c 1 192.168.1.2 || /root/resetlink.sh
2) resetlink.sh (if connected to port 1, adapt if you need)
uci set network.@switch_vlan[0].ports = '0 3 4'
uci commit network
/etc/init.d/network restart
sleep 1
uci set network.@switch_vlan[0].ports = '0 2 3 4'
uci commit network
/etc/init.d/network restart
It would be simpler to affect an IP address to the switch itself and use it for the ping target.
It should be possible to do the same using the swconfig utility, however I don't know the exact syntax to perform a port reset.
Regards,