Offshore Jacket Platform (OJP) constructions are often damaged by many factors such as earthquakes, tsunamis, and the effect of marine environment. Therefore, it is necessary to keep regard on the health of the construction to prevent deterioration and avoid damage to property and people. In this paper, we develop a data acquisition and processing system that uses a multi-sensor network combined with a deep learning neural network to identify anomalies for OJP due to the direct effects of environment such as waves, wind, and other direct impacts during mining. Noise-filtered data and two-dimensional scalogram rendering through a wavelet transform are used as input to train and test the designed system. The central processor STM32 Nucleo-F411RE considered the deep learning algorithm on an embedded model in MATLAB and the test results proved the effectiveness of the proposed method.