Design And Implementation Of Data Operation Module For Smart Pipe Gallery Cloud Platform

Driven by intelligence city policy,smart pipe galleries have received more attention as the underground core of intelligency city.Smart Pipe Gallery Cloud Plateform applies emerging technologies such as Internet of Things(IoT)and edge computing to realize intelligent management of underground pipe gallery,which effectively improving the comprehensive management level.Based on the Smart Pipe Gallery Cloud Plateform,this thesis has designed and implemented functions such as configurable buried points,log rule alarms,and abnormal log alarms.In order to effectively detect the condition of the Smart Pipe Gallery Cloud Plateform,this thesis proposed a design and implementation plan based on the data operation module of the Smart Pipe Gallery Cloud Plateform from three aspects:operation analysis,log management,and abnormal alarms.The main contents of this thesis are as follows:(1)Designed and implemented operation analysis module.To improve the understanding of the usagestatus of the Smart Pipe Gallery Cloud Plateform,this thesis proposed a user behavior monitoring scheme,which realizes the collection of user behavior data through the decorator syntactic and customized scripts,the pre-configured burried element information was a prerequisite for deciding whether to persist storage.Compared with common data collection and storage methods,this solution avoided the intrusion of business codes and a large amount of waste of resources.In the end,user behavior data combined with tools such as Echarts are visually presented in multiple dimensions,providing a digital basis for subsequent work.(2)Optimized the log management architecture.This thesis proposed a scheme based on Filebeat to implement log collection,which reduces CPU and memory usage.In order to strengthen the effective monitoring of the server which was dependenied by this plateform,this scheme also proposed a server performance detection scheme based on Metricbeat after considering the architecture of log system.And the performance status was shown by Kibana.Used ElastAlert to set alarm rules,by regularly querying the log changes in ElasticSearch,it can effectively monitor a large number of abnormal situations when the rule is triggered.It improved the detection efficiency of the plateform through sending alarm emails.(3)Designed and implemented the abnormal alarm module.This system might generate abnormal information which is identified usual such as excessive service startup time.Effective detection and differentiation of these information are the keys to improving system performance.Therefore,this thesis proposed an anomaly detection scheme based on the one-class SVM algorithm.By analyzing the log records,the rationality would be detected.And the maintenance staff would be notified in time when an abnormality occured,which reduced the platform risk during the operation and maintenance process and improved the system performance.