Design And Implementation Of Selection And Evaluation System For Electric Power Big Data

With the continuous development of Internet,human beings have entered the era of big data.The electric power industry complies with the requirements of the times,and uses power big data to enhance the value of the power industry.Machine learning method is an important tool for the rational use of power big data.In the application process of machine learning,the appropriate feature engineering depends on the understanding of business data by electric power professionals,and the appropriate machine learning algorithm and super parameter selection depend on the experience of data scientists.It is difficult to have both of them.Therefore,there are still difficulties in giving full play to the value of electric power big data,so practitioners need convenient tools to assist the application of machine learning algorithm in electric industry data.To solve this problem,this paper focuses on big data of electric power,studies and implements machine learning model evaluation and selection methods,provides the construction function of machine learning pipeline for classification and regression tasks,and provides simple data analysis and processing capabilities to assist data scientists in building and selecting models.Further,in order to facilitate the practical application of the above algorithms,this paper constructs the selection and evaluation system of large data algorithms for electric power,which is a one-stop intelligent machine learning model construction platform.The system encapsulates large data algorithm selection and evaluation algorithms for electric power.Electric power workers can obtain well-constructed machine learning algorithm configuration by simple clicks and conduct data mining independently.At the same time,the data scientist can use the system to analyze data quickly,build a good baseline model,and build algorithms to meet special requirements through personalized configuration.This paper filters out the core algorithm of large data of electric power by designing the evaluation standard of the core algorithm of large data of electric power.Then,the evaluation method and configuration selection method of the learning algorithm for large data machines in electric power are studied and designed.The former solves the problem of how to evaluate the performance of the machine learning configuration on a specific dataset,the latter receives the feedback from the former and chooses a high potential machine learning configuration within the optimal space.The two methods are combined to complete the automatic configuration optimization function of the learning algorithm for large data machines in electric power.Through tests,the above methods have good performance and efficiency,and run stably in the system.The system is designed and developed under the guidance of the principles of software engineering.Firstly,the background and significance of the research are introduced,and the research focus of the article is clarified based on the analysis of business status and related technologies.Then,typical business scenarios are analyzed,typical use cases are designed,and functional and non-functional requirements of the system are discovered.Subsequently,solutions are presented for the core issues that need to be solved in the system.Then,the overall design and detailed design of the system are carried out,the system architecture is designed,the function and interaction relationship of modules and sub-modules are clarified,and detailed introduction is made through UML class diagram and interaction diagram.Finally,the system is tested according to the requirements stated in the demand analysis to verify the validity of the system.Finally,the deployment and testing of the platform are explained,and the paper’s work is briefly summarized and prospected.