Monthly Statistical Line Loss Calculation Based On Data Mining And Association Rules

At present, the asynchronism of the meter reading cycle of power supply quantity and sale quantity makes line loss calculation results inconsistent with the actual situation and increase or decrease unrealistically, which made the statistical line loss calculation results cannot be compared with theoretical appraisal index. As a result, it is of great significance to solve the problem of the asynchronism of the statistical line loss calculation.This paper proposes a calculation model of the actual quantity sold based on Least Squares Support Vector Machines method(LSSVM), which effectively solved the problem of the sync power, making statistical line loss comparative with theoretical line loss and providing valuable reference for electric power company. The main contents are as follows:First of all, the basic concept and theory of data mining algorithms are introduced.The algorithm, LSSVM, which is suitable for the calculation of the interval quantity sold of two adjacent months is chosen. Meanwhile, association rules mining and its algorithm are introduced.Secondly, mine for the relationship between electricity-changing and random factors through the theory of association rules. Economic membership functions and weight coefficient are built, and the influence of variation of electricity on random factors is quantified for unit time, which modified the calculation results, making them more accurate and comprehensive. A monthly statistical line loss calculation method based on LSSVM and association rules mining is proposed. The information fusion theory is introduced to low pressure area users meter reading data information fusion, on the basis of equivalent meter example, the equivalent meter range can be got, which overcame uncertainty and complexity problem of the low voltage users.In the end, on the basis of preliminary results, economic factors and the result of association rule mining are used for multiple correction. So the monthly quantity sold over the same period of supply and monthly actual statistical line loss can be obtained. A limit statistical line loss calculation method based on association rules recommendation is alsoproposed. Through calculating the correlation coefficient and model recommendation, the using range of the original method is enlarged, and the practicability of the calculation is improved. The feasibility and practicability of this method are proved according to the simulation and analysis.