A new framework for power system identification based on an improved genetic algorithm

Accuracy parameters of system model are of great importance in stability and security evaluation or simulation for power system. Some of the conventional methods may have inadequate adaptability or effectiveness for identification of different power systems. A new framework for power system identification is proposed based on an improved genetic algorithm with a logarithmic fitness function and an adaptive search space. The framework can be used for most power system models (linear or nonlinear) and can easily be performed on different models just by rebuilding corresponding map lists between the system parameters and the model coefficients. The numerical experiment and practical experiment of a 600 MW steam turbine unit are conducted to examine the performance of the framework. The identification results have demonstrated the effectiveness of the proposed framework.