Abstract:Vibration data from physical systems, such as civil structures and machinery, often carries important information about the dynamic characteristics, but streaming acquisition of higher -frequency vibration often accrue large volumes of data, resulting in data transmission and storage challenges. Compressive sensing (CS) is a relatively newly-developed technique for efficient data representation, capable of reconstructing the target signal using only a few random measurements through sparse optimization. However, the real-world application of CS is hindered by the strong assumption of signal sparsity and a costly reconstruction process. In this work, we propose a novel deep learning method for vibration data reconstruction by using deep convolutional generative adversarial networks (DCGAN), which is composed of a generator G and a discriminator D. A modified 1D symmetric U-net architecture with shortcuts is presented for G to flexibly deal with different inputs, while a typical 1D classifier is used as D. A composite adversarial loss function is proposed considering errors in both time and frequency domains. The proposed DCGAN approach has several appealing properties. First, it directly learns the end-to-end mapping between the compressed and original signals without employing the sparsity assumption or random sampling, which fundamentally differs from existing sparsity -based CS methods. Second, the reconstruction process is highly computationally efficient as the network is fully feed-forward and no optimization is needed during data reconstruction. The proposed DCGAN approach is evaluated using the simulation data from a numerical 9-floor frame as well as experimental data collected from a large test steel grandstand. The results demonstrate the superiority of the proposed DCGAN in computational accuracy and efficiency compared to the tested sparsity-based algorithms. Furthermore, the influences of network configurations (network depth, down-sampling strategy, and shortcuts) are comprehensively explored.

Statistical Compressive Sensing Based on Convolutional Gaussian Mixture Model

Convolutional Gaussian Mixture Models with Application to Compressive Sensing.

Compressive sensing by learning a Gaussian mixture model from measurements

Efficient Compressed Sensing Reconstruction Using Group Sparse Total Variation Regularization

Task-Driven Adaptive Statistical Compressive Sensing of Gaussian Mixture Models

Image reconstruction based on back propagation learning in Compressed Sensing theory

Online Adaptive Statistical Compressed Sensing of Gaussian Mixture Models

A Fast Sparsity-Free Compressive Sensing Approach for Vibration Data Reconstruction Using Deep Convolutional GAN

Image compression based on Gaussian mixture model constrained using Markov random field

Compressive Sensing Image Coding with Perceptual Weighting Measuring Matrix

Image Reconstruction Based on Sub-Gaussian Random Projection

A combined HMM-PCNN model in the contourlet domain for image data compression

Statistical compressed sensing based on Bayesian principal component analysis

Block-Based Compressive Sensing Coding of Natural Images by Local Structural Measurement Matrix.

Learned Image Compression with Gaussian-Laplacian-Logistic Mixture Model and Concatenated Residual Modules

Image Encoding of Compressive Sensing Based on JND

Distributed Compressive Sensing: A Deep Learning Approach

Generalized Gaussian Model for Learned Image Compression

An efficient algorithm for compression-based compressed sensing

Group-based Sparse Representation for Image Compressive Sensing Reconstruction with Non-Convex Regularization.