(55co) Research on Leak Detection of Ethylene Spherical Tank Based on Convolution Neural Network

The leakage of ethylene spherical tank will pose a serious threat to safety production. However, it is difficult to detect the leakage in the initial stage. This paper studied on a method of automatic leakage detection using convolution neural network. It built a recognition model based on convolution neural network with the scene images of normal and leakage conditions from different angles. The model used maximum pooling method and ReLU activation function for optimize the network performance. The test results show that the proposed method and model can realize the automatic leakage detection effectively of ethylene spherical tank.