Study On Tariff Safety Rosk Early Warning Model Of Power Supply Enterprise

Tariff is power supply enterprise’s lifeline, directly affect the economic benefits and investment return of power supply enterprise, related to the power supply enterprise’s survival and development. So, strengthening the electric safety risk management, building charge of electricity safety risk early warning management system, effectively preventing electricity safety risk is a very important content.Based on the analysis of 《national grid company marketing safety risk prevention and management norms》 and researching of the risk early warning model of Each industry, the paper tries to call the SG186system’data, and then establish the electricity safety risk early-warning model, if the model can program through computer language, it will be able to automatic objective the tariff safety risk for early warning, so can save a lot of power supply company human, material and finance, and can implement real time monitor, the electricity safety risk.This article through analysising the electricity safety risk data type of power supply company’SG186marketing system, Classificate the electricity safety risk into two class--the safety risk and the electricity tariff recovery audit risk. And according to the specific situation of power supply company and two kinds of different risk characteristics, build the exponential smoothing method of tariff recovery safety risk early warning model and RBF neural network electricity audit security risk early warning modelwhich is based on the principal component analysis method. The exponential smoothing method of tariff recovery safety risk early-warning model use the the total arrears and user owes cost the total number to warning, in order to play for recovery of electric safety risk prevention function. Tariff audit security risk early warning model by KMO test to verify the electricity checking safety risk for principal component analysis, and principal component analysis method to select the total contribution rate of85%of the index, and the index established a RBF neural network prediction model. Finally, an example proves that the model is scientific and reasonable.