Low-voltage area prediction model research based on self-organizing competitive neural network

The distribution network voltage level is directly related to residents´ normal use of electricity, in order to correctly predict the low-voltage distribution network voltage quality, to take timely and effective means to prevent low-voltage phenomenon, the article collects nine indicators to reflect the characteristics of low-voltage distribution network voltage quality which are mainline diameter, mainline type, branch line diameter, branch line type, power supply radius, distribution transformer capacity -load ratio, the three-phase load unbalance rate, single-phase-home number, reactive power compensation rate. Then the article establishes a self-organizing competitive neural network model to automatic cluster the samples into three kinds which are normal, existing low voltage risk and existing severe low voltage risk. Using the known practical result to compare with the calculation results will indicate that the network model has high accuracy and feasibility.