Event classification for natural gas pipeline safety monitoring based on long short-term memory network and Adam algorithm

Abstract

Hydrate plugging and pipeline leak can impair the normal operation of natural gas pipeline and may lead to serious accidents. Since natural gas pipeline safety monitoring based on active acoustic excitation can detect and locate not only the two abnormal events but also normal components such as valves and pipeline elbows, recognition and classification of these events are of great importance to provide maintenance guidance for the pipeline operators and avoid false alarm. In this article, long short-term memory (LSTM) network is introduced and applied to classify detection signals of hydrate plugging, pipeline leak, and elbow. Adaptive moment estimation (Adam) algorithm is introduced and utilized to accelerate the long short-term memory network convergence in training. Experimental results demonstrate that the network with three layers and 64 units per cell performs the best. The cross-entropy loss in training is 0.0005, and classification accuracies are all 100% in training, validation, and testing which verify the validity of the long short-term memory network. Therefore, the method based on the long short-term memory network and adaptive moment estimation algorithm can work efficiently on pipeline events classification and has great guiding significance for safety assurance of natural gas transmission.

Gas pipeline event classification based on one-dimensional convolutional neural network

“…Therefore, the training process of the neural network is accelerated and the difficulty of events classification is reduced. The specific process of Z -score standardization is as follows 12…”

“…At the end of each iteration, 720 samples are shuffled, in which 540 samples are randomly re-selected as the training dataset and the rest 180 samples are the validation dataset for the next iteration. Since it can accelerate the loss reduction and avoid falling into local optimum of loss, adaptive moment estimation (Adam) algorithm 12 is applied to update the parameter weights w and bias b .…”

“…Therefore, long- and short-term memory (LSTM) network was applied to distinguish pipeline block, pipeline leak, and pipeline elbow. 12 The results of network training show that the trained LSTM network can achieve 100% classification accuracy of non-aliasing pipeline events. However, there are still some shortcomings in the LSTM network, such as long training time and the great fluctuation of training results.…”

A Machine Learning Method for the Detection of Brown Core in the Chinese Pear Variety Huangguan Using a MOS-Based E-Nose

“…The parameter update rule is defined as: where σ is the learning rate. As an extension of the stochastic gradient descent method [ 32 ], the Adam algorithm possesses fast speed in the optimization process without falling into the local optimum [ 33 ].…”

Deep wavelet sequence-based gated recurrent units for the prognosis of rotating machinery

“…Gated neural network, as a kind of recurrent neural network (RNN), is able to extract the dependencies hidden in the sequential data with controlling the previous information. 22 Compared to long short-term memory network which captures long-term information, 23 Gated recurrent unit (GRU) has simpler recurrent architecture but performs better on less amount of training data. To improve the prediction accuracy, the deep GRU architecture is modified with extra wavelet sequence layer, given the fact that the bearing data are heavily non-stationary and it is hard to capture the time-frequency features if we want to take advantage of the known physics.…”