Structure Analysis Of Shale And Parameter Prediction Based On Petrophysical Model And Neural Network

Shale is a common sedimentary rock.Unlike clastic rocks and carbonate rocks,shale has very small porosity and permeability.Early petroleum geological studies used shale as a good cover for oil and gas reservoirs,but With the rise of shale oil and gas and unconventional oil and gas exploration,shale is regarded as a very important unconventional oil and gas reservoir with great development potential.Shale is a multi-mineral heterogeneous mixture composed of quartz,feldspar,calcite and other minerals,clay,organic matter and other components.Nowadays,shale is usually regarded as a very important unconventional oil and gas reservoir.Compared with reservoirs,the occurrence of oil and gas in shale is complicated,the composition of rock minerals is diverse,the abundance of organic matter type varies greatly,and the pore space is complex and diverse.These characteristics determine that the interpretation and evaluation of shale oil and gas must be very different from conventional oil and gas reservoirs,so its evaluation ideas and research methods should be more suitable for the shale itself.Therefore,it is necessary to further research the relationship between the mineral composition,content and microstructure of the shale reservoir,and obtain accurate and valuable reservoir evaluation parameters.This article mainly analyzes shale reservoirs rich in organic matter,studies the structural characteristics of shale reservoirs and obtains shale reservoir parameters.The methods adopted include rock physics methods and applied neural network analysis.The petrophysical method is to establish a petrophysical model in the shale reservoir,and analyze the reservoir structure based on the model,and invert the physical properties of the reservoir.Petrophysical modeling is the most critical step in reservoir description and analysis.To establish a petrophysical model in shale reservoirs rich in organic matter,we must first study the influence of the structure and properties of organic matter(the main component is kerogen)on the shale.Therefore,we must survey comprehensively when establishing a petrophysical model of shale reservoir,the morphology,size,porosity and occupied space of the organic mixture are added to the rock based on the Voigt-Reuss-Hill model during the modeling process,which is approximately equivalent according to the medium self-consistent SCA(Self-consistent approximations),the theory quantitatively analyzes the effect of organic matter mixture occupying pore space and aspect ratio on the elastic properties of shale formations.The shear wave velocity calculated by the established rock physics model verifies the correctness of this model in the shale reservoir in the study area.The model rock physical inversion is then used to obtain shale reservoir evaluation parameters.Due to the rapid rise of research on big data,artificial intelligence,and deep learning in recent years,many researchers have applied intelligent algorithms such as neural network algorithms to the study of geophysical problems.The extraction of seismic attributes and the calculation of formation parameters involve a large amount of data computing,using intelligent algorithms such as neural networks can effectively solve these large amounts of data exchange problems.For example,the common BP neural network uses the algorithm of back propagation error.This algorithm is widely used in the simulation of nonlinear functions,but the BP neural network is easy to converge to the local minimum value and then jumping out of the network.Genetic algorithms are used to improve the BP neural network,and the algorithm is used to weaken the influence of the BP network's own defects,so that the new neural network can be better used for calculation,and prediction of organic shale reservoir parameters and seismic interpretation.