Research On Grid Short-Circuit Current Online Forecasting And Current Limiting Decision-making

In power system engineering design,calculating short circuit current is usually needed for selecting short circuit current limitation measures and setting protection devices.At present, the domestic main operation method of short-circuit current is the calculation curve method.The method can't completely meet the requirement of the calculation results accuracy and actual production needs,because of its limited calculation method. With the development of the computer, the domestic and foreign scholars have developed a series of large-scale simulation software for accurately calculating short circuit current at any time, in view of the defects of the traditional method.But the existence of disadvantages that collecting device parameters need a lot of work and creating transient model is complex and difficult at running time,and that it needs a lot of debugging, therefore these software are not suitable for practical engineering application engineering units.This paper analyses the causes of the short circuit current increasing and the shortage of monitoring and the current situation calculation analysis for short-circuit current.It studied the precision and rapidity of calculation at the same time, and the main contents are as follows:Starting from the basic model of power system, this paper analyzes the mathematical model of the electric power network, the parameters of the generator model and load model.The model selection combines the calculation accuracy and rapidity, and lays a theoretical basis for the following analysis of short-circuit current calculation, super short term forecasting and rapid calculation according to the measured.In order to improve the power grid short-circuit current calculation speed, considering that the power system node admittance matrix has the features of highly symmetric, sparse elements, quick and easy creation and modification, the fault current calculation method based on sparse matrix transform is proposed. At first, the admittance matrix generation process had been optimized; Secondly, triangle decomposition method was applied to the admittance matrix decomposition to form factor table for quick calculating; At last,based on the low order update principle of sparse matrix and improved non-negative matrix decomposition method, the low order operation of the matrix was used to replace the original admittance matrix modification; Due to the low order update operation with the uniqueness of solution,it can quickly get accurate short circuit current value.Based on the short circuit current calculation results of sparse matrix transform, in order to more accurately predict the future moment of short-circuit current level, first of all, thesimilar day load multipoint extrapolation method had been used to derived and predict trends within five minutes of the current moment load values. Secondly,the improved genetic algorithm optimize neural network training model was used to iterative train normalized input and output sample values. Finally, the future time load as the last set of input samples,was foretasted into the trained model to output ideal value of short-term prediction circuit, reaching for the purpose of network scanning, analysis and early warning in advance.On the basis of super short term forecasting short circuit current, we need to determine whether the current value of each node after five minutes in a safe state,combining with the breaking capacity of protection devices. When the predicted short-circuit current may be excessive, limiting decisions can been timely and automatically calculated based on the sparse matrix transformation method,to select the appropriate decisions of breaking or putting into operation,in order to achieve the policy to meet short-circuit current limiting effect.At last, through the simulation result of the IEEE system, a complete set of short circuit current calculation implementation scheme was getted,which contains "calculation analysis--forecast scanning--decision-making and control" of the online.On the other hand, the validity and practicability of the presented method was verified,and it provides a theoretical basis and specific guidance for engineering decisions.