Research And Implementation Of Earthquake Precursor Data Analysis And Anomaly Monitoring

Earthquake disasters seriously threaten people's property and safety.The improvement of disaster early warning capabilities can avoid huge losses caused by earthquakes and enable people to live a stable life.Earthquake disasters occur frequently in various countries in the world.Since 1980,my country has counted more than 200 earthquake disaster research reports.Among them,more than 3/4 of the areas have an earthquake intensity of more than five,covering most of my country's land territory.Natural disasters will severely damage our economy.The economic losses caused by the Wenchuan earthquake in 2008 exceeded 200 billion yuan.China is located in the border zone of multiple plates.Due to the severe plate activity,earthquake disasters occur frequently in the southwestern China and in the coastal zone.Since the 21 st century,there have been many serious earthquakes.The suddenness and unknown nature of earthquake disasters It will also cause huge casualties.The disasters caused by earthquakes are serious,and earthquake early warning work is particularly important.This article focuses on the extraction of anomaly information in earthquakes.The precursory data of earthquakes are closely related to earthquakes.The data is diverse,with wide sources,strong analyzability,and anomalies The data is an important basis for judging the occurrence of earthquake disasters.The equipment that collects precursor data can realize regular collection and interval collection.The precursor data is in the process of continuous update and change,so it is necessary to monitor the precursor data and find the appropriate anomalies.Data detection algorithm.This paper has completed the anomaly monitoring algorithm research and earthquake precursor data analysis.The specific contributions are:(1)Completed data preprocessing.The clustering algorithm uses tagging to process seismic precursor data.The traditional data preprocessing process only has normalization operations,and the seismic precursor data is unlabeled data.The lack of labels will cause support vector machine(SVM)classification Inaccurate.Data cleaning is completed,and the data cleaning operation will remove the erroneous data that interferes with the operation of the algorithm.(2)Completed improved PSO(Partical Swarm Optimization,PSO)combined with support vector machine to detect precursory anomaly data.Use the Ito lemma to optimize the PSO initialization process,improve the particle mutation ability,and make the initial particles diverse.The Zigguart algorithm is used to increase the number of PSO particles to solve the problem of imbalance between the global search ability and the local search ability of the system.The adaptive speed update method is used to improve the speed and position update formula of particles in the PSO algorithm.The improved PSO combined with support vector machine method is used to implement the application of seismic precursory data in anomaly detection algorithms.The actual precursory data samples collected by Chuzhou Seismological Bureau are detected by the algorithm in this paper,and the anomaly data obtained are real and effective.(3)Established a platform for monitoring and early warning of earthquake precursor data.Using the B / S architecture system based on JAVA language,data analysis,online data collection,database data access,SMS contact person in charge and data display functions are completed,and the anomaly detection system is used as a part of the earthquake warning data warning platform.In summary,the improved anomaly detection algorithm improves the anomaly recognition rate and prediction rate.Compared with the PSO algorithm,MOPSO / DC algorithm,and PSO-A algorithm,the convergence of the algorithm in this paper is the global search ability,the stability of the operation is the best,the algorithm has a high fitness value and good fitting effect;the precursor data passes through the algorithm of this paper Compared with the PSO-SVM algorithm,SVM algorithm,and GWO-SVM algorithm after SVM classification,the effective anomaly data recognition accuracy rate and recognition accuracy rate are the best.The anomaly detection algorithm for precursor data can effectively extract abnormal data.