Logging Response Characteristics And Lithology Identification Of Volcanic Rocks In Huoshiling Formation,Southern Songliao Basin

As the largest Mesozoic continental petroliferous basin in China,the Songliao Basin has attracted the attention of scholars at home and abroad for its abundant resources.In recent years,major exploration breakthroughs have been continuously made in the volcanic rock strata of the Huoshiling Formation in the southern Songliao Basin,which indicates that the volcanic reservoirs in the southern Songliao Basin have good prospects for exploration.However,as the degree of exploration continues to deepen,the basic understanding of the genetic mechanism of volcanic rock reservoirs and the formation of reservoir space is insufficient,which has brought certain obstacles to exploration.To solve the above problems,it is first necessary to clarify the lithology of the volcanic rocks of the Huoshiling Formation.However,due to the limitation of the source of core cuttings samples,only geological samples cannot be used to construct lithological sequences of the entire well.Relatively speaking,conventional logging data respond obviously to changes in volcanic rock lithology,with strong vertical continuity and high resolution.Therefore,it is very important to use logging data to achieve lithological sequence reconstruction.This paper takes 18 wells drilled into the Huoshiling Formation in the Changling fault depression and Lishu fault depression in the southern Songliao Basin as the research object.Through core description and thin section identification,the main development types of the Huoshiling Formation volcanic rocks are determined,and these lithological characteristics are summarized.Logging response characteristics and extract their logging data.Based on the logging data of these volcanic rocks,the cross plot and the BP neural network that has been widely used in various fields in recent years are used to predict the lithology of the volcanic rocks in the study area.The reconstruction of the volcanic rock lithology sequence of the Huoshiling Formation lays the foundation for subsequent volcanic reservoir evaluation.1.The main types of volcanic rocks in the Huoshiling Formation in the southern Songliao BasinThrough core description and thin section identification of the core section of the Huoshiling Formation in the Changling fault depression and the Lishu fault depression303 m,it is determined that there are 8 types of volcanic rocks in the Huoshiling Formation in the southern Songliao Basin,namely basalt,andesite,and andesite.Quality breccia lava,andesitic breccia,andesitic tuff,rhyolite,rhyolite breccia and rhyolite tuff.2.Logging response characteristics of volcanic rocks in the Huoshiling Formation in the southern Songliao BasinBy analyzing the logging curves of 8 volcanic rocks developed in the study area,the logging response characteristics of different volcanic rocks are summarized.The natural gamma value of the volcanic rocks in the study area gradually increased from basic rocks to acid rocks.Among the andesites,clastic rocks were larger than those of lava.In terms of compensation neutron values,basalts were the highest;among the sonic time difference values,basalt is the lowest,andesitic breccia lava and rhyolite breccia are higher than other lithology.Rhyolitic breccia has the lowest deep and shallow lateral resistivity value;the values of other lithologies are significantly higher than that of rhyolitic breccia;each lithology has an approximate density value,but the density value of rhyolitic breccia is lower the magnitude of change is minimal.3.Lithology identification of volcanic rocks based on conventional logging data and comparison and optimizationA total of 258 sets of logging curve data for eight types of volcanic rocks are selected and randomly divided into training data and prediction data at a ratio of 7:3.The training data is used to establish the lithological prediction model,and the prediction data is used to calculate the coincidence rate of the lithological prediction model.In the method,the conventional cross plot and the BP neural network are selected.The cross plot predicts that the coincidence rate of the volcanic rocks in the study area is up to 55%;the BP neural network is used,and Dropout is introduced to prevent overfitting.The results show that the introduction of Dropout The prediction coincidence rate of the BP neural network for volcanic rocks in the study area reached 89%,which can effectively distinguish the main volcanic rock types in the study area.4.Methodology applicationCarried out in-situ training in-situ application and in-situ training and remote application.This model is used to predict the entire well interval of the Huoshiling Formation volcanic rocks in 2 wells in the Changling fault depression.After comparing with cores and cuttings,it is found that the lithological prediction model is applicable in the study area.At the same time,the core section of the drilling in the Hailar Basin is selected,and the corresponding logging data is input into the BP neural network,and the output result is compared with the original lithology to evaluate the cross-regional applicability of the model.The results show that the neural network trained with the original training set cannot effectively distinguish the volcanic rocks in the Hailar Basin.Therefore,a total of 592 sets of standard logging data of 8 volcanic rocks in the Hailar core section were selected,and the original BP neural network structure was maintained under the premise of targeted training,the coincidence rate is reached 91.7%.