The multilayer perceptron (MLP) network has been widely used in various fields. However, the performance of neural networks can be heavily affected by the choice of the parameters, such as the number of hidden nodes. Here we propose a new scheme for finding an “optimal” architecture of a multilayer perceptron network for a given problem. Our philosophy is to associate a tunable gate with every node in the hidden layer. Initially, all gates remain almost closed indicating a scenario where all hidden nodes are practically rejected (i.e., non-existent). Then during training, the gates are opened depending on the ability of the nodes in reducing the objective function. In order to reduce the use of redundant (unnecessary) hidden nodes, we use an L 1 regularization on the extent of gate opening. Typically, the network begins with a large number of hidden nodes and after training, not-useful hidden nodes are deleted. The network then may further be tuned to adapt itself with the reduced architecture. The performance of the pruning method is tested on several datasets and compared with the Weight Decay and Weight Elimination regularizer methods. The results demonstrate the advantages of the proposed method on both pruning efficiency and generalization.
When the depressurization development of a hydrate-bearing layer is initiated, the temperature of the near-wellbore formation quickly decreases to near the equilibrium temperature due to the dissociation of natural gas hydrate Therefore, the secondary generation of natural gas hydrates in the wellbore easily occurs if pressure jumps to a high value due to the changes of production rates or shutdown of the well. Though hydrate generation in the process of high-pressure drilling and gas and water transportation has been widely investigated, the secondary generation of natural gas hydrates caused by pressure jump during the depressurization development process is not fully understood. In this study, the multiphase pipe flow of a horizontal well, the VyniauskasBishnoi generation dynamics of natural gas hydrate, and a decomposition dynamics model developed by Kim and Bishnoi are combined to build a set of horizontal well gasliquidsolid three-phase flow models, which consider the phase transition in the wellbore and distinguished the secondary hydrate generation area in the wellbore under different temperature and pressure conditions. The results show that when the toe-end pressure is 7MPa and the environment temperature is 6.4C, the secondary hydrate generation exists in the horizontal section of the horizontal well, and the maximum hydrate flow velocity in the wellbore is 0.044m3/d. A high toe-end pressure, low environment temperature, and high gas output will result in a greater hydrate generation in the wellbore, and the wellbore pressure will have a remarkable influence on the amount generated and its range.