The mean,median and mode are statistical means often used in statistics to describe the concentrative trend of a group of data.However,the three kinds of data are not only different in definition,but also different in...The mean,median and mode are statistical means often used in statistics to describe the concentrative trend of a group of data.However,the three kinds of data are not only different in definition,but also different in application.For the same group of data,there is no definite and strict standard for which one should be used.What's more,different data may even lead to diametrically opposite conclusions.Even if we finally choose one,what extent the data can describe the concentrative trend of the group of data?Is there any statistical method that can describe the concentrative trend of a group of data more objectively and fairly?It is not only a very interesting issue,but also an issue including great research value.展开更多
基于经验模式分解方法和长短期记忆网络(empirical model decomposition and long short-term memory network,EMD-LSTM)模型对水位数据进行预测。先采用中值滤波对数据序列进行预处理,然后对数据序列进行EMD分解,并对EMD分解的每个特...基于经验模式分解方法和长短期记忆网络(empirical model decomposition and long short-term memory network,EMD-LSTM)模型对水位数据进行预测。先采用中值滤波对数据序列进行预处理,然后对数据序列进行EMD分解,并对EMD分解的每个特征序列使用LSTM模型进行预测,最后叠加各个序列预测值,得到最终的预测结果。以南水北调工程某河流每隔1 h的瞬时流量、流速和水深监测数据为研究对象,采用EMD-LSTM模型进行建模,试验结果表明,该模型能够实现水位、水速和瞬时流量连续12 h和6 h的准确预测,且比LSTM模型具有更高的预测精度,可为水位预判和水资源的实时调度提供决策依据。展开更多
为了解决基于霍尔传感器混合动力汽车防抱死系统轮速检测信号容易产生温度漂移干扰的问题,提出了一种利用联合中值均值加权和经验模函数分解估计温度漂移干扰信号的算法.通过联合中值均值加权估计出温度漂移趋势成分后,再对估计温度漂...为了解决基于霍尔传感器混合动力汽车防抱死系统轮速检测信号容易产生温度漂移干扰的问题,提出了一种利用联合中值均值加权和经验模函数分解估计温度漂移干扰信号的算法.通过联合中值均值加权估计出温度漂移趋势成分后,再对估计温度漂移趋势进行自适应固态模函数分解,利用t检验的方法,判断出各阶固态模函数中不属于温度漂移趋势的成分,继而得到温度漂移趋势的精确估计.对比了不同温度漂移干扰下本文算法与形态学滤波算法的噪声修正性能,结果表明,本文算法能够有效剔除温度漂移干扰,平均信噪比提升4 d B以上.展开更多
文摘The mean,median and mode are statistical means often used in statistics to describe the concentrative trend of a group of data.However,the three kinds of data are not only different in definition,but also different in application.For the same group of data,there is no definite and strict standard for which one should be used.What's more,different data may even lead to diametrically opposite conclusions.Even if we finally choose one,what extent the data can describe the concentrative trend of the group of data?Is there any statistical method that can describe the concentrative trend of a group of data more objectively and fairly?It is not only a very interesting issue,but also an issue including great research value.
文摘基于经验模式分解方法和长短期记忆网络(empirical model decomposition and long short-term memory network,EMD-LSTM)模型对水位数据进行预测。先采用中值滤波对数据序列进行预处理,然后对数据序列进行EMD分解,并对EMD分解的每个特征序列使用LSTM模型进行预测,最后叠加各个序列预测值,得到最终的预测结果。以南水北调工程某河流每隔1 h的瞬时流量、流速和水深监测数据为研究对象,采用EMD-LSTM模型进行建模,试验结果表明,该模型能够实现水位、水速和瞬时流量连续12 h和6 h的准确预测,且比LSTM模型具有更高的预测精度,可为水位预判和水资源的实时调度提供决策依据。
文摘为了解决基于霍尔传感器混合动力汽车防抱死系统轮速检测信号容易产生温度漂移干扰的问题,提出了一种利用联合中值均值加权和经验模函数分解估计温度漂移干扰信号的算法.通过联合中值均值加权估计出温度漂移趋势成分后,再对估计温度漂移趋势进行自适应固态模函数分解,利用t检验的方法,判断出各阶固态模函数中不属于温度漂移趋势的成分,继而得到温度漂移趋势的精确估计.对比了不同温度漂移干扰下本文算法与形态学滤波算法的噪声修正性能,结果表明,本文算法能够有效剔除温度漂移干扰,平均信噪比提升4 d B以上.
