Research And Implementation Of Data Intelligent Analysis System Under The Industrial Internet Platform Of Power Generation Enterprises

In recent years,with the development of China’s power generation enterprises,power generation has increased rapidly year by year.However,continuous rise of production costs and the weak innovation ability of power generation enterprises have become more prominent.Therefore,how to improve the informatization and intelligence level,reduce management costs,and improve production efficiency are key issues to enhance the core competitiveness of China’s power generation industry.In order to realize informatized and intelligent industrial production,more and more power generation enterprises tend to build industrial Internet platforms to fully interconnect people,machines,and goods.They analyze a large number of data generated in daily production,thus providing intelligent support for corporate development decision-making.However,most mainstream industrial Internet platform solutions in the power industry,such as Proficloud and ABB Ability,cannot provide customized data modeling and analysis functions well.Further,if the existing data mining software is directly integrated into the industrial Internet platform,there are still some problems,such as low integration with the characteristics of power generation enterprises and high user threshold.In response to above problems,based on Syncplant 5.0 platform developed by Seiyon Automation Group Co.,Ltd.,this thesis studies and implements a set of data intelligent analysis system under the industrial Internet platform of power generation enterprises.The specific work includes the following three aspects:Research typical machine learning algorithms for platform users,and design machine learning support services that support dynamic configuration.Through the integration of multiple mainstream machine learning algorithms covering regression,classification and clustering,and intelligent recommendation of excellent features and parameters,a one-stop model customization function is realized.This study greatly reduces the technical threshold of machine learning application for platform user,while maintaining data intelligent analysis and dynamic prediction.In addition,practical models are preset for typical application scenarios in the power generation industry,so as to provide users with excellent data mining experience in the industry.To verify the usability of machine learning support services,this thesis,orienting towards platform management,designs an application service of user behavior anomaly detection based on user data access.Firstly,the user’s data operation characteristics are extracted from syntax,environment and return results to construct user behavior mode.Then,BP neural network,C4.5decision tree and random forest are used to train and verify the data sets of different sizes.Through the comparison of accuracy,false positive rate and false negative rate,C4.5 decision tree is finally selected to build the model of user behavior anomaly detection,and good experimental results are obtained.Design and implement the data intelligent analysis system under the industrial Internet platform of power generation enterprises.Through test and verification,the system is ensured to effectively improve the intelligent analysis ability of the industrial Internet platform of power enterprises.To sum up,based on Syncplant 5.0 platform developed by Seiyon Automation Group Co.,Ltd.,this thesis studies and implements a set of data intelligent analysis system under the industrial Internet platform of power generation enterprises,including machine learning module for platform users and user abnormal behavior detection module for platform management.It also realizes the intelligent analysis,dynamic prediction,and security management of data,thereby effectively improving the intelligent analysis ability of the industrial Internet platform of electric enterprises.