Wireless communication network anomaly detection method based on approximate dependency rule

The invention relates to the technical field of intelligent operation and maintenance, and particularly provides a wireless communication network anomaly detection method based on an approximate dependency rule, and the method comprises the following steps: based on network operation and maintenance data and key feature index category information, efficiently constructing an operation and maintenance probability graph for describing the dependency relationship between key feature indexes and the uncertainty of the dependency relationship; approximate function dependency mining is realized according to credible mutual information among key feature indexes, an operation and maintenance probability graph structure is used for pruning redundant dependency, and a score upper bound is used for pruning low-score dependency, so that mining efficiency is improved; selecting an approximate function dependency sensitive to the network state abnormal sample as an approximate dependency rule; and performing anomaly detect