In the O&M field, operators encounter the following common problem if there is a fault: large number of alarms, difficult fault locating, and numerous invalid dispatches. An operator in country J is used as an example. The number of reported alarms differs greatly from the number of valuable dispatches within one month, and 99.96% of reported alarms are invalid.
Use the AI training platform to learn alarm correlation rules based on historical alarms and topology database data. The real-time alarm sequences identify flag alarms based on alarm correlation rules and help implement cross-domain alarm correlation, greatly reducing the number of duplicate dispatches. Analyze fault root causes based on flag alarms, the fault cause library, and maintenance database, reducing invalid site visits caused by faults, such as software faults.
Operator M is used as an example. A total number of 2 million alarms are reported on a day. After the alarms are merged in the centralized fault handling system, most alarms can be filtered out, and the dispatch system generates about 4000 dispatches. The system can filter out 30% of duplicate dispatches and 30% of invalid dispatches based on the analysis and the AI accurate dispatch solution. The system can further filter out 30% of unnecessary site visit dispatches based on root cause analysis. Finally, there are about 1400 valuable dispatches.