Research On Correlation Analysis And Fusion Method Of Power Network Planning Index Based On Probabilistic Box Theory

Network planning is an extremely important part of the entire power grid system in the early planning process, the quality of the system is running, economic and social decisive part, and the difficulty of planning increased because of many uncertainties in power system planning. Since there is a certain relationship between many certain indicators in the inter-network planning(there is a strong correlation between a client voltage passing rates and power supply reliability), if ignored this relationship on the one hand it will bring the computational complexity of the problem, may bring serious curse of dimensionality; on the other hand can cause an error in the calculation of the project, thereby affecting the accuracy of decision-making, and may even cause some economic losses. To solve the above problem, This paper focused on the relevance of network planning model, the relationship between the technical evaluation network planning were discussed in detail, and modeling solving the uncertainty in network planning,studying the correlation of parameters and indicators of relevance based on the theory of probability and using indicators of relevance fusion algorithm. it propose the index correlation analysis and fusion algorithm based on the theory of the probability boxes.Firstly, study on technical evaluation theory and indicators in the network planning which discusses various single indicators and the specific meaning of the formula, and correlation analysis with multi-source single indicator data, After obtain the corresponding correlation coefficient, using the coefficient as a weight to make fusion process, screened a greater impact on the overall assessment of the indicators and to establish the probability index box;Secondly, According to the characteristics of the various indicators, based on probability theory box and write the corresponding algorithm program with Matlab procedures,for multi-data,using Copula models to obtain the correlation coefficient between the index and reference Lucas model’s interval feature, finally get the interval of the correlation coefficient after N sampling times;Finally, each indicators has established a good probability boxes,and analysis it with the correlation coefficient interval which improved weighted fusion method.Models and algorithms proposed above is applied to the case study,a provincial network planning indicators were evaluated, little effect on the overall index can be filtered in the pre-treatment metrics, thereby greatly reducing the workload; when fulfill full positive correlation or strong correlation between the indicators, knowing that one of the index value can be used instead of the known indicators and metrics related to sexual gratification known positive correlation or strong correlation another indicator. Such network planning for future work in the modeling process provides a better guidance method, especially for certain indicators under unknown circumstances, can find the indicators relevant indicator instead of full positive correlation or strong correlation between the indicators, the paper will compare traditional correlation method with based on probability theory Copula correlation cartridge models on the correlation process, the results show that the proposed method in this paper can fully contains characteristics of all the uncertainty indicators, significantly better than not treatment on the efficiency and effectiveness, the results showed correlation model or paper gives fusion algorithm is effective.