由日内高频收益率数据计算得到的已实现波动会受到市场微观结构噪音的影响和干扰，进而会对资产价格波动率的估计造成偏差。与此同时，市场中出现的异常价格跳跃行为也会使已实现波动呈现出特殊的统计特征。因此，已实现研究市场微观结构噪声及价格跳跃对已实现波动的影响，对准确刻画资产价格波动率具有重要意义。本文首先对多种已实现波动测度进行检验，用中国股票市场数据检验了不同已实现波动测度对价格跳跃情况的敏感程度，并对实证检验出的价格跳跃现象进行了分析。其次，通过建立基于列维稳定分布的新的高频资产价格生成机制，并控制噪声强度、跳跃幅度等会影响分布特性的隐含波动参数，模拟生成不同市场情形下的价格序列，并估计出相应的已实现波动及其长记忆性。最后，文章将模拟实验的结果和实证分析相结合，对抽样频率、波动序列的厚尾程度以及噪声强度等因素与已实现波动长记忆性的影响进行探讨，并研究了价格跳跃行为与已实现波动的长记忆性之间的关系。模拟实验的结果表明，已实现波动序列的长记忆性会随着抽样频率的增大以及噪声强度的增加而显著增强；价格波动序列的厚尾程度并不是造成已实现波动出现长记忆性的主要因素；价格跳跃行为会对已实现波动序列的长记忆性产生不同程度的影响，尤其在噪声强度较大的情况下，已实现波动序列的长记忆性会受到跳跃频率和跳跃幅度较为显著的影响。

The impact of market microstructure noise can affect the realized volatility, and thus cause the deviation of the estimated volatility. This influence becomes much more significant with the increase of sampling frequency. Therefore, the problem of how to improve the realized volatility estimation method and to reduce the effects of market microstructure noise has become a wide concern in the study of high frequency financial data.This paper first does a series of tests on several kinds of realized variance and their sensitivity to jumps of asset prices. An empirical study on asset prices jump is also conducted based on samples from Chinese stock market. Then, artificial price series is generated to simulate market price in different situations. In the generating mechanism, log return is set to fit stable levy distribution, the parameters of extent of noise, jump intensity, jump amplitude are introduced to control the price process. Then different factors that may affect the long memory of realized volatility, including the sampling frequency, the degree of volatility sequences of thick tail and noise intensity are analyzed based on the simulation.The simulation results show that the long memory of realized volatility series increases significantly with the increase of sampling frequency and the noise intensity, the extent of the fat does not necessarily cause the long memory realized volatility, volatile prices jump impacts the long memory property of realized volatility series in different extents, especially when the intensity of noise becomes stronger, the long memory of realized volatility series is significantly affected by the frequency and the amplitude of jumping.