Research On Parallel Architecture Of Energy Management System Based On OpenTSDB

The energy management system is an important tool for energy management and production monitoring in manufacturing industries,especially high-energy-consuming industries.It provides theoretical and data support for enterprises to formulate energysaving and emission-reduction measures.The frequency of energy data collection determines the fine-grained granularity of data analysis by energy management systems.Efficient query analysis speed can help enterprise managers to detect the occurrence of energy consumption anomalies in a timely manner and take corresponding countermeasures.The traditional single-point relational database shows obvious performance shortcomings in practical applications.It is an inevitable trend to combine big data technology with energy management system.In this thesis,Foshan's aluminum industry encountered difficulties in highfrequency acquisition and storage in the process of energy management system,and the historical energy data query analysis was slow to respond to two problems.The corresponding big data solution was proposed.The specific research work is as follows:1)Understand the overall structure and logic of the energy management system,according to the table structure and fields in the SqlServer database currently used by a certain aluminum industry in Foshan,analyze the energy consumption data;and estimate the high-frequency acquisition according to the current data volume in the database.The amount of data.2)Designing a heterogeneous database storage scheme for the problem of highfrequency acquisition of energy data storage blockage loss,the original SqlServer database is used to store basic data,high-frequency acquisition,rapid growth of energy data,and distributed storage system To store,and through the comparative analysis,select the HSS cluster-based time series database OpenTSDB,use OpenTSDB high throughput,support the multi-TSD node parallel write characteristics,make full use of the cluster processing power,obtain high-speed write capability,support The energy data is collected in the second order and effectively solves the energy data storage problem of an aluminum industry in Foshan.3)The analysis of energy data is slow,and the Spark parallel computing framework is used to process the massive energy data stored in the cluster.In order to improve the efficiency of Spark computing,a consistent hash algorithm is used to load balance the energy data between nodes,and the Compressed Bloom Filter algorithm is used to filter the irrelevant data,thereby reducing the network transmission cost between nodes,thereby improving the speed of the associated join operation.The speed of energy data query analysis has been effectively improved.Based on the above research,from the high-frequency acquisition and storage of energy data to the rapid query and analysis of massive energy data,the idea of parallel computing of big data is integrated into the energy management system.In the end,the experimental comparison test was carried out.Compared with the existing storage query scheme,the performance improvement is obvious,which can meet the highfrequency second-level energy collection data of an aluminum industry in Foshan,and the rapid response requirement of query analysis.