基于进化计算的类噪声负荷辨识改进方法

This paper proposes a non-simulation model identification model different from the ambient signal load models, and successfully applies it to the daily load data identification of a power system. It is defined as the Hammerstein-Wiener model from the perspective of system identification, and is improved and ensured in terms of the model structure, the algorithm initial value strategy, the differential evolution algorithm operator and the basic properties. The Hammerstein-Wiener model is significantly improved in its accuracy and convergence speed, and it is demonstrated that there are no numerical problems that often occur in the previous model optimizations. After a multi-perspective analysis, it is verified that this complete and effective set of analysis is applicable to the ambient signal load identification, a great significance for load modeling.