Study On The Neural Network Method For Oil-gas Bed Recognition

Recognizing the oil-gas-water bed exactly is both one of the critical link of improving the effect of exploration and offering important basic data for the disposition and planning of development. Especially in the medium-late period of water-injection development oil field, the content of water of production in oil well is more and more high, so knowing exactly the situation of the undergrad oil-gas-water bed is very necessary. The civil well log company get the suit rate of artificial interpretation for oil-gas bed is though very high, it mainly depended on the rich experience of the log engineers, and it is very disadvantageous for the popularization of the oil-gas-water bed interpretation. And the present software for interpretation is not including the artificial experience; the suit rate of computer interpretation and the artificial interpretation have a great difference.This title based on the present recognition study and evaluation technique, through the analysis for the logging response in different reservoir, taking out the characteristic parameter of well log cove, and use neural network to recognize reservoir, it is believed that the suit rate of reservoir log interpretation with computer will get a good improvement. The research of this title, for the exploration and development of oil field and oil pool, for stabilizing oil and controlling water, and for the high productivity and stable production of oil field, will have a very important guiding meaning.Concrete content and expected result:1. Give the normal interpretation for log data of the oil field developing area, mark the standard reservoir.2. Considering plenty the effect factor of reservoir, make a statistics of the log curve response character of reservoir; establish a suitable recognition pattern of standard reservoir.3. Take out the characteristic parameter of the log curve for reservoir; make sure that the effect factors of different reservoir reflect on the characteristic parameter.4. Choose the neural network method, join with the characteristic parameter which taken out from the log curve, write a recognition programmed of reservoir neural network, and make a treatment of practical log data, to raise the suit rate of computer recognition for reservoir.