Hi!
The problem is following: you generate event SYS_THRESHOLD_REACHED when you cross threshold. This event has sverity WARNING. Then, you have the rule in event processing policy, which has two filtering conditions: if event is either SYS_THRESHOLD_REACHED or DC_HIGH_CPU_UTIL AND event's severity is CRITICAL, then generate alarm. Your event will never pass through this filter, because it has severity WARNING. Also, your custom event - CPU_USAGE - is not used at all.
You should do consistent config, for example as following:
In threshold, set event to DC_HIGH_CPU_UTIL
In event processing policy, leave "severity" column as "any", and leave only DC_HIGH_CPU_UTIL in "event" column
Optionally, change alarm text to %m, if you wish to have alarm text to be the same as event's message
Best regards,
Victor