Application of Forward-Backward Confidence Estimation Fusion Filter in Real-Time Landslide Monitoring

Real-time Global Navigation Satellite System (GNSS) monitoring of landslides plays a crucial role in disaster prevention and the protection of lives and property. However, noise in monitoring data directly affects the accuracy of disaster early warning systems. Specifically, under high-amplitude signal fluctuations, existing data filtering techniques often exhibit suboptimal performance and slow convergence, making it challenging to effectively mitigate the impact of noise. To address this issue, this paper proposes a Forward-Backward Confidence Estimation Fusion Filter (FBCEFF) algorithm. Based on confidence estimation filtering, the algorithm integrates forward and backward processing to comprehensively analyze data trends within the sliding window, accurately identify anomalous fluctuation periods, and refine results to produce the final filtered sequence. Experimental results from measured data demonstrate that the proposed algorithm excels at suppressing high-amplitude fluctuations, significantly improves data accuracy, and achieves faster convergence compared to traditional unidirectional filtering methods.