The Research Based On Radial Basis Function Neural Network In Voltage Stability

In the power system operation process, there are many random disturbance and uncertainty factors, So voltage stability has been the main threat to the operation of many power system, However, because of its bad influence to society, voltage collapse accident can cause great economic loss, it becomes an emergent problem to be solved for the security of power system.The margin of the voltage stability as a important basis can be intuitive to diagnose voltage stability of the system, In order to get the margin of the voltage stability of each node of that system accurately, and put forward a kind of RBF neural network algorithm, it has the convergence of high precision, fast convergence rate and strong fitting ability, etc. While the standard BP algorithm often lead to local minimum values, and has a disadvantages of slow convergence speed, so put the of method additional momentum and the method of adaptive learning rate link up an improved BP algorithm, namely the two algorithm is put forward in this paper.Adopting two kinds of models that simulated each node’s data of IEEE5system, it takes the simulation results and plot the PV curve of each node of the system. By using the limit value of voltage stability that is predicted, it calculates the margin of the stability of each node, which can judge the weak point in the system.Through the comparison of the errors is produced by the above two methods, we can judge that the RBF neural network is more accurate and fast to work out the voltage stability limit of the system than BP network, Finally, it takes the system of IEEE30node for example, using RBF neural network to simulation. The feasibility has yet been proved.