Farzad Delnavaz - Tehran Islamic Azad University Science and Research ,Tehran, Iran
Jamshid Moghadasi - Department of Petroleum Engineering, Petroleum University of Technology, Ahwaz, Iran
Mohsen Masihi - Tehran Islamic Azad University Science and Research ,Tehran, Iran
Mohammad Javad Nabavi-Zadeh - Tehran Islamic Azad University Science and Research ,Tehran, Iran

Analysis of well logs to obtain petrophysical data from hydrocarbon reservoirs is very important. This process is a complete set of log-scale storage needs, for various reasons suchas lack The chart fully in old wells,the loss of logs on the effects of improper maintenance, falling some distance, which the logs are incomplete the log analysis is not done , or itcan be done with low precision .Therefore in this study was to try NPHI logs artificialwells P1,P2, P3, P4 in the South Pars gas field using artificial neural networks to be it. The assessment or collection of this log which logs to predict the influence of a log andhas the least impact and finally obtain the optimal network structure. The three-layer perceptron network with back propagation learning rule is error. Sigmoid function of stimulus in the middle layer and output layer is linear function