Fault Line Selection And Location In Distribution Network Based On Improved PSO-BP Algorithm

The single-phase ground fault occurs most in the distribution network grounding fault,the symmetry of the three-phase voltage has not been destroyed in a short time,but the long run will lead to further expansion of the fault,the development of two-phase or even three-phase short circuit,causing system over-voltage.Therefore,it is important to judge the fault line quickly and accurately and determine the location of the fault point.As the fault transient and steady state signal is affected by the grounding resistance,the phase,the neutral point operation mode and the field electromagnetic interference,which leads to a single line selection and ranging method is limited.In order to overcome the limitation of single steady state or transient selection and distance measurement method,an improved particle swarm optimization(PSO)optimized back propagation(BP)neural network is proposed to select the line and distance measurement method.Aiming at the shortcomings of BP algorithm,such as slow convergence and susceptibility to initial weights,the PSO algorithm has the advantages of fast convergence,strong local and global optimization performance to make up for the defects of BP algorithm,In order to prevent the PSO algorithm from converging in the late stage and being easy to fall into the local optimum,using a comprehensive inertia weight and acceleration constant improvement strategy,and through the test function validation.The improved PSO-BP algorithm is applied to fault line selection in distribution network,Through the wavelet packet transform(WPT)to extract the transient component combined with the fifth harmonic component,the zero sequence active component as the input variable of the neural network,using the improved PSO-BP algorithm to complete the training and testing,output line selection results.The simulation results show that,compared with BP algorithm,PSO-BP algorithm,improved PSO-BP algorithm have fast convergence and high precision,and are not affected by factors such as grounding resistance,fault phase and fault location.On the basis of completing the selection line,the improved PSO-BP algorithm is applied to fault location.The wavelet energy,zero sequence reactive component and zero sequence fundamental component are extracted by WPT and fast Fourier transform(FFT)as the fault feature quantity,Considering the change of grounding resistance and fault phase angle,the training sample set is established.The training and test results show that the improved PSO optimization BP algorithm has better convergence speed and ranging accuracy than BP algorithm.