研究工作基于分形理论,应用G-P算法,通过对常州某股份有限公司事故时间序列的分析,构建了n维相空间,并利用关联维数的基本原理,对事故时间序列进行了分形特征分析.研究结果表明当嵌入维数达到10以后,该事故动力学系统具有稳定的关联维数4.1,说明至少有4个因子在影响着事故时间序列的动态变化,并且该系统的有效自由度为10.研究对建立事故时间序列的预测模型有较大的参考价值. Based on the fractal theory and the GP algorithm, this paper constructs the n-dimensional phase space by analyzing the accident time series of a limited liability company in Changzhou, and uses the basic principle of the correlation dimension to analyze the fractal characteristics of the accident time series. The results show that when the embedding dimension reaches 10, the dynamic system of the accident has a stable correlation dimension of 4.1, indicating that at least four factors affect the dynamic change of the accident time series and the effective degree of freedom of the system is 10. Research It has great reference value for establishing the forecasting model of accident time series.