Time series Bayesian compression sampling and signal decompression reconstruction method and data loss recovery method

Abstract

The present invention proposes time-series Bayesian compression sampling and signal decompression reconstruction method and data loss recovery method, said method includes the selection of measurement matrix for signal compression sampling, the design of basis vector matrix, the signal itself and its adjacent time period Modeling of variation sparsity, Bayesian probability solution for decompressed and reconstructed signals at each time period, fast optimal estimation of hyperparameters, diagnosis of reconstruction accuracy based on posterior uncertainty quantification, signal loss in wireless sensing for health monitoring recovery method, etc. The method of the present invention adopts the hierarchical sparse Bayesian learning modeling and solving algorithm, in the embedded signal itself and its two sparsity characteristics that change with time, the hyperparameters are quickly solved, and the signal reconstruction is robust under a higher compression rate It has unique advantages in terms of accuracy and uncertainty quantification of signal reconstruction, and it is also robust to noise.

Description

technical field [0001] The invention belongs to the technical field of signal processing and structural health monitoring, and in particular relates to a time-series Bayesian compressed sampling and signal decompression reconstruction method and a data loss recovery method. Background technique [0002] At present, under the joint action of factors such as environmental erosion, material aging, long-term load effects, and component defects, cumulative damage will inevitably occur in infrastructure, resulting in continuous reduction of its bearing capacity, continuous loss of functions, and even endangering the safe use of the structure. Therefore, ensuring the service safety and life of major projects is a major issue related to the national economy and people's livelihood. Structural health monitoring can effectively guarantee the safety, life and function of major projects by deploying large-scale, distributed sensor networks and data acquisition, transmission, management ...

AI Technical Summary

Problems solved by technology

[0006] At present, the research and application of Bayesian compression sampling in time series and image signals of structural health monitoring, in the robustness of the algorithm, the utilization of the structural characteristics of the signal itself, the processing ability and uncertainty of the time series signals acquired continuously over time There are still some deficiencies in aspects such as quantitative quantization, which make it difficult to meet the signal decompression accuracy of compressed sampling at a higher compression rate and difficult to maintain stability, and the superiority of compressed sampling technology has not been fully explored

Images