Research on recognition of precursor anomaly data based on improved reverse selection algorithm

In order to solve the problems of low accuracy of seismic precursor anomaly data detection method and long-term reliance on manual observation and calculation to identify precursor anomalous data, this paper proposes a precursor anomaly data identification method with improved Reverse Selection Algorithm, which uses the seismic precursor data provided by the China Earthquake Network as the input index, and uses the particle swarm optimization PSO optimization Reverse Selection Algorithm to train the dimensionally reduced data, so as to achieve the purpose of identifying abnormal data. It solves the problem that the traditional Reverse Selection Algorithm can only be set artificially and cannot independently determine the self-radius Rs, and the algorithm anomaly recognition rate is improved. Compared with other anomaly data detection algorithms such as traditional Reverse Selection Algorithm and support vector machine algorithm, the running speed, global search ability, stability of operation, and prediction rate of abnormal data recognition rate of the system are improved.