The Research Of Dynamic Monitoring Data Compression Technology Of Railway Bridges

With the vigorous development of the transportation, the proportion of bridge in the railway line keeps growing, playing an increasingly important role. However, the safety of the bridge gradually aroused people’s wide attention with the increasing of the number of bridges, span, operation time as well as the load, so the appear of bridge health monitoring systems have become an inevitable trend. With the establishment and installation of the system, the bridge administrator can know well the operation and safety state of the bridge, finding the security hidden danger in time and reduce the happening of the accident. However, due to the long time of the monitoring system, a large amounts of data ha ve produced, then it will bring more trouble to the transmission and storage of data. Therefore, research on the compression processing of the monitoring data become the inevitable trend of the development in order to solve the problem of the structural health monitoring, such as the limit of the bandwidth and save the storage space.In this article, some analysis methods and reconstruction algorithm based on the theory of compression sampling for monitoring data of railway bridge was proposed aiming at the magnanimity of the monitoring data, considering the dynamic load and environmental factors, and combined with the periodic and cyclestability of the data at the same time.Due to the immobility of the train and the consistency of the vehicle parameters running on the railway bridges, the monitoring data will changed with periodic. So in this article, first to verify the existence of the periodic through the simulation of the ANSYS, then to explore the internal periodical of monitoring data by the methods o f wavelet transform analysis and cyclic spectrum analysis, making preparation for subsequent analysis.Be different from the traditional sampling theory, compressed sensing is a new theory, which is not confined to the bandwidth limit during the process of sampling. Based on the characteristics of the monitoring data, this article makes thorough analysis about the principle of compressed sensing, and then comparing the advantages and disadvantages of various reconstruction algorithm, at last, the OMP algorithm was taken to the compression and reconstruction of the monitoring data, achieving satisfactory results, and proved the feasibility of the algorithm.Sparse Bayesian Learning is an emerging research field in recent years, and with the combination of Compressive sensing, it has a better reconstruction effect for block sparse signal. At the same time, this algorithm overcome the shortage of the OMP algorithm such as the complex of computing and the long time of iteration. Due to the periodic change of monitoring signal, this article takes the difference algebra to the data to make it having block sparse features, so that it can take advantage of the sparse Bayesian learning to compressed and reconstructed, and also it achieve satisfactory results. Otherwise in order to make the compressed and reconstructed process more simple and effective, the compressed sensing algorithm based on sparse Bayesian learning has taken in this article to compress and reconstruct the original acceleration signal(without sparse), and the results show that the algorithm is feasible and practical for nonsparse signal and can get better results.