为了得到更加精确的混沌时间序列信号的关联维数计算值, 提出了识别无标度区间的新方法. 首先基于K-means 算法将关联积分双对数曲线中明显不属于无标度区间的点剔除掉 其次提出点斜率误差算法从上一步保留下来的点中识别出一个大概的无标度区间 最后再次基于K-means 算法从大概的无标度区间中识别出更加准确的无标度区-To obtain more precise time sequence of signals 关联维 really counted, the identification of no mark on the new method is based on riverboats. first, the means are associated with the curve will be points of no mark clearly not part of the interval of time　then return to a slope from the errors are a step down in order to identify a glimpse of the marked degree. again, the algorithm based on riverboats means that no one from the interval of more accurate to identify the area is marked