变形监测中获得的观测数据不可避免地存在着噪声,如何从受噪声干扰的数据序列中提取特征信息,提高变形监测的精度是变形监测系统所涉及的关键技术问题之一。在赵瑞珍的阈值算法基础上,提出了一种改进的阈值算法,进而给出改进的模平方处理法。理论分析、工程实例的结果表明,改进的模平方处理法能够更有效地克服软阈值方法恒定偏差的缺点,进一步提高了消噪的精度和可靠性。 One of the key technical problems involved in deformation monitoring system is how to extract the characteristic information from the data sequence disturbed by noise and improve the accuracy of deformation monitoring. Based on Zhao Ruizhen’s threshold algorithm, an improved threshold algorithm is proposed, and then an improved method of modular squaring is given. Theoretical analysis and engineering examples show that the improved modular squared method can more effectively overcome the shortcoming of constant deviation of soft threshold method and further improve the accuracy and reliability of noise elimination.