Data Driven Efficiency Evaluation And Resource Allocation Of Power Grid Enterprise

With the rapid development of the electricity economy,the demand for electricity will con-tinue to grow rigidly in the future.However,the traditional power grid is faced with problems such as the old grid and lagging development,and the grid line loss is exacerbated.The power genera-tion efficiency is continuously low,and even power quality problems including voltage fluctuation and harmonic pollution are caused,which seriously affect the economic benefits of power grid en-terprises.Therefore,in order to maintain a strong competitive advantage at the new power and energy market in the future,power grid enterprises must reassess all aspects of grid production and construction,optimize and upgrade grid structure and improve grid production efficiency.In the meantime,the development of the power industry can not be separated from the positive interactionwith the regional economy.Therefore,the power grid investment decision-making process needs to accurately grasp the trend of electricity economy,and mine the potential information of the power industry data to assist enterprises in data-based decision-making.The arrival of big data era is both an opportunity and a challenge for traditional power grid enterprises.In order to ensure the sus-tainable and efficient production and operation,power grid enterprises urgently need to set up an effective efficiency evaluation system.Through in-depth study on production investment data,they need to refine customized investment planning strategies to improve the input-output efficiency of enterprises.In summary,the research work of this thesis mainly consists of the following three aspects:First of all,this thesis designs a set of power prosperity index method to analyze the trend of regional economic change,and describes the whole design from the establishment of data index library,data preprocessing,index screening to final index synthesis.In the whole process,the feature information of power data closely related to the trend of economic change is deeply tapped.Finally,the effectiveness of the index method in the provincial power economic analysis is verified,which provides a theoretical basis for regional investment decision of power grid enterprises.Secondly,in order to effectively analyze the weakness in the development of power grid,a fuzzy clustering algorithm based on the power prosperity theory is introduced to improve the eval-uation process of power grid enterprises.The power economic environment variables are taken as feature inputs to ensure the comparability of all evaluation units.In addition,based on the eval-uation theory of classical data envelopment analysis and the drawbacks of traditional algorithms,this thesis designs an improved evaluation model suitable both for bi-directional multi-input and multi-output.At the same time,an improved classification and simplification evaluation algorithm is proposed in combination with the existing data classification algorithms innovatively.Finally,the effectiveness of the algorithm is verified by experiments.It is concluded that average outage time and loss are the main factors that affect the efficiency of power grid enterprises.Finally,in order to improve the average power outage time and loss caused by benefit loss,this thesis studies in-depth study on the allocation of reactive power resources in power grid up-grades,and optimizes the site-based strategy of reactive power equipment in the case of the main line failure.In this thesis,we first establish the index of the electrical betweenness of the grid lines,and select the dominant line in the power flow through the electrical order.Secondly,considering the economic constraints-of reactive power compensation in the power grid enterprise,we estab-lished the optimization problem of the minimum generation cost,the maximum loss of networkloss and the optimal margin improvement under the to find out the optimal compensation strategy.Finally,the improved catastrophic genetic algorithm is used to solve the nonlinear mixed integer programming problem,and the algorithm is validated in standard system of IEEE 14 nodes.