开采沉陷监测数据处理的分段卡尔曼滤波模型

为解决长周期、高频率的GNSS煤矿开采沉陷监测数据受外界环境影响大、干扰噪声高、可靠性低的问题，采用卡尔曼滤波对开采沉陷GNSS监测数据进行分段滤波处理。首先采用回归分析方法将监测数据分割为开始阶段-活跃阶段-衰退阶段三部分。开始阶段、衰退阶段下沉较为稳定，以标准卡尔曼滤波模型对沉陷数据进行滤波处理；活跃阶段沉陷变化快，通过构建加入修正数的卡尔曼滤波模型对沉陷数据进行处理。采用Python语言建立滤波程序对宁夏某矿5年沉降监测数据计算分析结果显示，对不同阶段采用不同的卡尔曼滤波结果与实测过程曲线吻合，滤波效果良好。加入修正数的卡尔曼滤波模型可以有效地处理沉陷变化大的矿区监测数据，减少数据波动对沉陷结果的影响，提高监测数据的可靠性。研究成果为长周期、高频率的沉陷监测数据处理提供了科学依据。

Subsection Kalman filter model for mining subsidence monitoring data processing

Kalman filtering is used to filter GNSS mining subsidence monitoring data in order to solve the problems of long period and high frequency GNSS mining subsidence monitoring data that are greatly affected by the external environment, high data interference noise and low data reliability. Firstly, the regression analysis method is adopted to automatically divide the monitoring data into three parts: the initial stage, the active stage and the decline stage. The subsidence in the initial stage and the decline stage is relatively stable, and the subsidence data is filtered by the standard Kalman filtering model. The add correction kalman filter model was constructed to deal with the data in the active stage of rapid subsidence change. The filter program was established by Python language, and the monitoring data of five-year and hourly interval sampling rate in a mining area in Ningxia were calculated and analyzed. The results showed that the process curve of different kalman filtering results in different stages was consistent with the measured results, and the filtering effect was good. The add correction kalman filter model can effectively process the monitoring data of mining area with large subsidence variation. The method constructed in this paper can effectively reduce the impact of data fluctuation on the subsidence result and improve the reliability of monitoring data. The research results provide a scientific basis for long-term and high-frequency settlement monitoring data processing.