The invention relates to the technical field of
mobile communication network operation and maintenance, and discloses an automatic fault monitoring method and
system for
base station operation and maintenance. The method comprises the following steps: acquiring operation indexes of a
base station and performing
time sequence analysis to obtain a drift vector; according to the drift vector, track clustering is carried out by adopting a K-means
algorithm, and an initial performance baseline is determined to construct a
baseline model; deviation is calculated and filtering
processing is carried out to obtain a pure residual error sequence; micro degradation
modes in the pure residual sequence are identified and aggregated, and an early
signal list is obtained; constructing a correlation graph in combination with historical alarm logs, and mining a dependency relationship to obtain a correlation alarm cluster; performing
root cause positioning rule fusion and
feature aggregation on the cluster to obtain a labeled fault description; and finally, matching the early warning template and determining the priority to obtain final early warning output. According to the method, tiny hidden dangers of
base station performance can be effectively identified, the problem that massive alarms are difficult to associate is solved, and early-stage accurate early warning and
root cause positioning of faults are achieved.