中文摘要：

针对CB模型及其改进模型中由于规则网络描述的均质群体结构与真实金融市场中投资者相互作用的异质性相悖，提出基于表征异质投资群体结构的无标度网络的自组织金融模型．通过投资者在交易规则约束下的自组织聚簇行为，模拟金融市场的动态演化过程．模型生成的价格波动序列与实际股指具有相似的演化动力学：价格收益的概率分布具有尖峰胖尾的特征，且它的中心峰值与时间尺度存在幂律关系，这表明价格波动序列的演化是一个自相似过程；易变性具有明显的聚簇行为，说明价格波动序列具有连续的巨幅涨落和长程关联性．这些统计特性与金融市场实证相符，验证了模型的有效性．

英文摘要：

For a series of grid-based Cont-Bouchaud （CB） models unable to correctly represent the heterogeneity of interactions among investors in the real financial market, an improved evolutionary model constrained by trading rules was proposed based on the percolation theory on scale-free networks. The time series of price fluctuations generated by the model was similar to the stock index in the real financial markets. For instances, the probability distributions of returns showed the sharp peak and fat tail, and their peak values restricted to the time scales obey the power-law behavior, which suggests that the time series of price fluctuations evolves in a self-similarity way. The clustering behavior of volatility shows that there are large fluctuations and long-range correlations in the evolutionary process. These statistical properties of return and volatility empirically are consistent with the real financial markets, indicating the effectiveness of the improved model.