Research And Application Of Sonic Reconstruction In NP-4 Tectonic Zone

Well logging constrained seismic inversion is the core content of reservoir prediction technology, and for seismic inversion, analysis and application of logging data is very important. When the inversion results can effectively response the differences between the reservoir and wall rock, the seismic inversion results can accurately describe the reservoir.NP-4 tectonic area mainly developed fan delta sedimentary system, and has good exploration potential. But in the base of the fan, mud and sand flow together, in the front of the fan, thin layer deposition of sand and shale is developed, which leads to the sonic of sand and shale are difficult to distinguish, so the difficulty of the reservoir prediction is increased. If directly using the raw well logs to do the inversion, we unable to characterize reservoir accurately. This paper does the sonic reconstruction research aimed at the actual conditions of NP-4 tectonic area, in order to improve the sensitivity to the reservoir of sonic logs.The multi-logs sonic reconstruction technique can use gamma ray, spontaneous-potential, resistivity and neutron porosity to reconstruct sonic logs according to respective weighting coefficient and keep the original properties of raw sonic logs at the same time. This technique does not have to separate frequency like traditional technique, so it not only retains all frequency components of raw sonic logs, but also adds additional useful information of reservoir coming from other well logs, so the reconstructed sonic can better highlight the characteristics of the reservoir.Petrophysics modeling uses Gassmann, Xu – White model equation to calculate the elastic parameters of strata in the study area on the basis of study of petrophysics theory, then do the petrophysics forward of every wells’ density and sonic in NP-4 tectonic area. The forward logs can effectively highlight the characteristics of reservoir, and increase the sensitivity to the reservoir compared with the raw logs.Particle swarm optimized BP neural network reconstruction technology introduces the particle swarm algorithm to the weights adjustment process of traditional BP neural network, and optimize the BP neural network. The reconstruct sonic logs using optimized neural network has better ability of distinguishing reservoir compared with the raw logs, and solves the problem that the training of traditional BP neural network can be easily paralyzed and has poor effect.In this paper, according to the characteristic of different sonic reconstruction techniques and actual additions of wells in the study area, selecting the best reconstructed sonic logs to do the well log constrained sparse pulse inversion, the results shows that the reconstructed logs can significantly improve the effect of impedance inversion and accuracy of reservoir prediction. Therefore the above-mentioned sonic reconstruction scheme proposed by this paper can satisfy the needs of reconstructing sonic logs in NP-4 tectonic area.