فهرست مطالب زهره بیات زاده

Summary Horizontal directional drilling (HDD) is a popular method for installation of both steel and plastic underground pipelines. Besides selecting the appropriate type and size of reamers the rotational torque is another important parameters that must be predicted for performing the reaming operation. In this study, hybrid artificial neural networks (ANN) with biogeography-based optimization (ANN-BBO) model were applied for predicting rotational torque. In fact BBO was used to better regulate the weights and biases of the ANN model. In this study, axial force on the cutter/bit, rotational speed of the bit, the length of drill string in the borehole, the total angular change of the borehole, the radius for the ith reaming operation, the mud flow rate and the mud viscosity are applied as input variables to predict the rotational torque. To assess the ability of the model to predict the rotational torque, West–East Natural Gas Transmission project in China was used. Results indicate that this model has high potentials for estimating the rotational torque using a set of listed input parameters.
Introduction A major concern of many HDD projects is prediction of required rotation torque. It has been established that the required rotational torque at the drill rig depends on various factors, including geological conditions, drilling method, reamer cutter/bit size and type, rotary speed, axial force on bit, drilling mud properties, borehole diameter, length of drill string in the borehole, and borehole trajectory. In this area in recent years, studies have been done using traditional statistical methods, but this study focuses on the application of artificial intelligence in this field.
Methodology and Approaches In this study, ANN method and BBO algorithm is used. We used BBO to better regulate the weights and biases of the ANN model. BBO is an evolutionary algorithm that is inspired by biogeography. In BBO, a biogeography habitat indicates a candidate optimization problem solution, and it is comprised of a set of features, which are also called decision variables, or independent variables. BBO consists of two main steps: migration and mutation.
Results and Conclusions In this paper, 75% of the data sets were assigned for training purposes while 25% was used for testing of the network performance. Network with 7-8-1 structure is optimized and the results indicate that this model has high potentials for estimating the rotational torque.