Intelligent Computing and Innovation on Data Science by Unknown

Author:Unknown
Language: eng
Format: epub
ISBN: 9789811532849
Publisher: Springer Singapore

In this stage, feed-forward network structure is used, shown in Fig. 2. In the input layer, six nodes are present which represents six input parameters (PV, YP, dh, dod, up and mud weight), whereas in output layer, only one neuron representing surge pressure. MATLAB educational package is used to develop the intelligent model. In neural network, backpropagation algorithm was used for network building and training. For the development of backpropagation network, total number of hidden layers with nodes had to be defined. Addition to this, several other parameters like training function, transfer function, adaption learning function and number of learning iterations also need to be selected. Due to lack of literature available in this topic, trial-n-error process is applied to select the most appropriate combination of above given parameter. The best combination was log-sigmoid transfer function, scaled conjugate gradient backpropagation (training function), gradient descent with momentum weight, bias learning function and 100,000 number of iterations. The structure of the neural network changes according to the complexity of the sample data. In this work, 1–3 hidden layer was varied with each layer consists of neuron in the range of 2–16. Total of 108 networks with variations of hidden layer and number of neurons were developed and trained.

Fig. 2Schematic diagram of feed-forward ANN structure

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.