A One-Dimensional Signal Data Restoration Method Based on Convolutional Neural Network

Abstract

The invention discloses a method for repairing one-dimensional signal data based on a convolutional neural network. First, the damaged data is subjected to boundary processing, and a convolutional neural network model based on an Encoder-decoder architecture is constructed. The loss in the convolutional neural network model is The weighted processing of the function makes the repaired data more instantaneous. The method of the present invention can restore all the lost data by fitting the damaged signal without any prior knowledge, and solve the problem of wireless transmission. The problem of data packet loss caused by factors such as link instability and node failure in the sensing network; adding L2 regularization items to avoid over-fitting phenomena in the network, and selecting the double conditional limit of the number of iterations and the rate of change of the loss function as the stop condition for training, has It contributes to the stability of the performance of the neural network and improves the efficiency of data restoration; the Adam optimization algorithm is adopted, which has fast convergence speed and good anti-noise performance, and can correct the convolutional neural network model more quickly.

Description

technical field [0001] The invention belongs to the technical field of data processing and data restoration, and in particular relates to a one-dimensional signal data restoration method based on a convolutional neural network. Background technique [0002] Due to node failure, wireless communication packet loss, and data loss often occur during the transmission process of the wireless sensor test network. [0003] Common one-dimensional data repair algorithms include: adjacent interpolation algorithm, EM algorithm, CS algorithm. Among them, the typical adjacent interpolation algorithm is segmented interpolation algorithm, which is often applied in the scene of slow-changing signal restoration with obvious temporal and spatial correlation, such as environmental temperature and humidity test, and is not suitable for signal restoration with high-frequency changes; EM algorithm is suitable for Statistical data repair is not suitable for application scenarios with few samples a...

AI Technical Summary

Problems solved by technology

In practical applications, the time correlation of signals monitored by wireless sensor networks is not necessarily obvious, and even there is no spatial correlation, so it is difficult to judge its sparse basis based on prior knowledge.

Images