Abstract: To find out the internal dynamic characteristics of economics fluctuation, the possibility of the existence of chaotic behavior in a city's power consumption fluctuation is analyzed nonlinearly. Based on the city's data logging of power consumption in recent years, the different methods are used to analyze the chaotic dynamics, such as the phase space reconstruction, amplitude spectra and Poincare's mapping. The correlated fractal dimension calculated. As a result, the saturated embedded dimensions are 7 and the correlation fractal dimension is 0.834, and a judgment is thus primarily made that the process of power consumption in the city is chaotic. The dynamically chaotic characteristics of power consumption fluctuation is therefore proved rigorously, thus providing a theoretical basis for studying the relevant forecasting of chaos.