Reliability of distribution network based on entity extraction model of characteristic spectrum

The structure of the distribution network will change with the continuous access of distributed generation, which will affect the distribution of power flow in the original system network. When a fault occurs in the distribution network, it will cause the short-circuit current of the protection equipment on the transmission line to change, so the analysis of the distribution network and power system becomes more complex. This paper introduces the entity extraction model of the characteristic spectrum to analyze the causes of prediction errors in the distribution network. The probability distribution model of prediction errors is established. Feature extraction is carried out on the processed samples. The weights of different input influence features are adjusted and optimized through algorithm optimization. Entity extraction is carried out while quantifying and comparing the risk levels to identify the types and levels of faults. The simulation results show that the neural network combination algorithm model has achieved excellent prediction accuracy and operation speed. It is proved that the method of mechanism-data combination can realize early visual warning. The algorithm system studied in this paper can scientifically evaluate the situation of the distribution network and ensure the level of stable operation of the distribution network.