论文部分内容阅读
针对模糊近似熵方法在生成时间序列数据特征过程中出现的依赖参数较多和计算复杂度较高的问题,提出了相关近似熵方法,并应用在传感网数据故障检测中.相关近似熵方法采用相关信息熵来计算向量空间中多维数据之间的相关度,通过计算向量空间在其维数由M维增加到M+1维时多维数据之间保持相关性的概率来判定一个时间序列的复杂程度.相对于模糊近似熵,相关近似熵方法将依赖参数从4个减少到了2个,并减小了计算复杂度.实验结果表明:相关近似熵生成的特征在大多数情况下显著优于模糊近似熵生成的特征,并且相关近似熵方法大幅度地缩短了传感器数据特征的生成时间.
Aiming at the problems that the fuzzy approximate entropy method has more dependent parameters and higher computational complexity in the process of generating time series data features, a relative approximate entropy method is proposed and applied in the fault detection of the sensor network data. The related information entropy is used to calculate the correlation between multidimensional data in vector space. By calculating the probability that vector space maintains the correlation between multidimensional data when its dimension increases from M to M + 1, Relative to the fuzzy approximate entropy, the relative approximate entropy method reduces the dependent parameters from 4 to 2, and reduces the computational complexity.The experimental results show that the features generated by the relevant approximate entropy are significantly better than most of the cases The features of fuzzy approximate entropy generation are blurred, and the related approximate entropy method shortens the generation time of the sensor data features.