The invention provides a wind power random fluctuation based generalized load modeling method which comprises the steps of: (1) determining a generalized load model structure, wherein the generalized load model structure comprises a reactive compensation device, a static load, an induction motor load, an asynchronous generator, a power injection module and equivalent impedance of a power distribution network; (2) obtaining aggregation parameters of the asynchronous generator with a statistical synthesis method, and obtaining electric parameters of an equivalent asynchronous generator by performing capacity weighting aggregation on each asynchronous generator; (3) identifying parameters of a load model; (4) configuring a user-defined program and generating a dynamic link library; and (5) injecting power fluctuation data, and during transient analysis and small disturbance characteristic value analysis, shielding the power injection module and only considering dynamic characteristics of the asynchronous generator. When wind power random fluctuation based characteristics are subjected to medium and long-term dynamic analysis, the power injection module is activated for injecting power to a system in a time-varying current form.