我国原油供应的对外依存度已达 70%，严重威胁能源安全，是目前亟待解决的“卡脖子”问题。我国页岩油气储量居世界首位，但由于页岩主要以陆相沉积和中低成熟度为主，现有的非常规油气开发技术无法高效应用。因此，理解页岩油气产生机理具有重大战略意义。干酪根是页岩油气的生成母质，具有分子量大、结构复杂、性质多样等特点，对它的演化机制研究是理解其产油产气机理的关键。干酪根经过早期成岩作用之后，由于埋藏温度的升高，进一步演化受热裂解的影响，因此，理解干酪根热演化的量子力学机制是认识干酪根形成、实现页岩油革命的关键之一。干酪根成熟度与地质演化阶段直接相关，是判别油气生成潜力的重要依据，有助于进一步预测干酪根裂解生成油气产量。

基于以上背景，本文从力学的角度出发，结合理论、实验、机器学习等方法，研究干酪根的热演化的量子力学机制。目前干酪根成熟度主要通过实验测量，其中最常用的指标是镜质体反射率。然而镜质体反射率并不能反映出干酪根成熟度变化过程中化学结构的演变规律，因此，基于量子力学与轨道杂化，提出了一种新的干酪根成熟度指标，轨道杂化成熟度指数 (Orbital Hybridization Maturity Index, OrbHMI)。该指标进一步阐明了干酪根熟化过程中化学结构变化的底层物理机制，OrbHMI 以 sp² 和 sp³ 杂化碳的比例来表征干酪根热解过程中化学键的断裂和重组，可以更深层次地解释干酪根的热演化机制。当 sp² 杂化碳比例升高时，OrbHMI 的值也随之升高，干酪根成熟度增加，分子结构向芳香结构转变。由于轨道杂化是量子层面的参数，而成熟度属于干酪根的宏观性质，利用传统研究方法费时又费力，因此，引入机器学习技术来阐明轨道杂化与成熟度之间的关系。结合大量核磁共振数据构建了基于轨道杂化的干酪根成熟度表征模型，成熟度的平均预测误差仅为 4.91%，超过 87% 的测试样本误差小于 10%。结果表明该模型能够准确预测干酪根的成熟度。

基于流体力学和量子力学，推导了不可压缩薛定谔方程的一般形式，退化后的方程与不可压缩薛定谔流控制方程相符。基于此结论与机器学习方法，建立了预测不可压缩薛定谔流波函数的机器学习模型，比较了模型参数的影响并确定了最终参数。得到的结果在预测波函数上表现良好，对双组分波函数的预测总误差低于 5%。

本文研究了干酪根的热演化的量子力学机制，建立了基于机器学习的干酪根成熟度表征模型，探究了干酪根熟化过程化学结构的演变，提出了基于轨道杂化的成熟度指标。这些研究为认识人工催熟、预测干酪根结构和生烃潜力提供新思路。同时分析了流体动力学方程的量子力学形式，建立了神经网络模型用于预测不可压缩薛定谔流的波函数。

The external dependence rate on crude oil supply of China has reached 70%, which seriously threatens energy security and has become the key bottleneck. Although the quantity of China’s shale oil and gas reserves ranks the first in the world, the existing unconventional oil and gas development technologies are not applicable because the shale is mainly composed of continental sediments with medium and low maturity. Therefore, understanding the mechanism of shale oil and gas production is of great strategic significance. Kerogen is the parent material for gas-generating and oil-forming. It has the characteristics of large molecular weight, complex structure, and various properties. Therefore, studying the evolution mechanism is the key to understanding the mechanism of oil and gas production. After the early diagenesis, the further evolution of kerogen is affected by thermal cracking due to the increase in burial temperature. Therefore, understanding the quantum mechanical mechanism of kerogen’s thermal evolution is one of the keys to recognizing kerogen formation and achieving the shale oil revolution. Kerogen maturity is directly related to the stage of geological evolution. It is one of the crucial parameters to determine the potential for hydrocarbon generation, which is helpful to predict oil and gas yield from kerogen cracking.

Based on the above background, this thesis studies the quantum mechanical mechanism of kerogen’s thermal evolution from the mechanics perspective, combining theory, experiment, and machine learning methods. Currently, kerogen maturity is mainly measured by experiments, and the most commonly used indicator is vitrinite reflectance. However, the vitrinite reflectance cannot reflect the change in the chemical structure during the evolution of kerogen maturity. Therefore, a new kerogen maturity index (Orbital Hybridization Maturity Index, OrbHMI) is proposed based on quantum mechanics and orbital hybridization. This indicator further clarifies the quantum mechanical mechanism of chemical structure changes during kerogen maturation. OrbHMI adopts the ratio of sp² and sp³ hybridized carbons to characterize the breaking and recombination of chemical bonds during kerogen pyrolysis, which can be applied further to understand the mechanism of kerogen’s thermal evolution. It is found that as the proportion of sp² hybridized carbons increases, the value of OrbHMI increases, which means the kerogen maturity increases during the evolution of maturation, and the molecular structure towards the aromatic structure. Since orbital hybridization is from the quantum level, and maturity is a macroscopic property of kerogen, it is time-consuming and labor-intensive to use traditional research methods. Therefore, machine learning techniques are introduced to clarify the relationship between orbital hybridization and maturity. Combined with a large number of nuclear magnetic resonance data, a characterization model of kerogen maturity based on orbital hybridization is constructed. The average error of the predicted values is merely 4.91%, and more than 87% of the test samples have an error of less than 10%. The results demonstrate that the model can accurately predict kerogen maturity.

The general form of the incompressible Schrödinger equation is derived based on fluid mechanics and quantum mechanics. The degenerate equation agrees with the governing equation of incompressible Schrödinger flow. Subsequently, the machine learning model for predicting the wave functions in the incompressible Schrödinger flow is established. The influence of parameters is compared and determined in the model. The results show good performance in predicting the wave function, and the prediction error of the two-component wave function is less than 5%.

In this thesis, the quantum mechanical mechanism of the thermal evolution of kerogen is studied, and the evolution of chemical structure during kerogen maturation is explored. Then a machine learning model for predicting kerogen maturity is established, and a maturity index based on orbital hybridization is proposed. The results provide new method for the research of artificial maturation, the kerogen structure, and the hydrocarbon generation potential. Furthermore, the quantum mechanical forms of the hydrodynamics equations are analyzed. Finally, the neural network model is established to predict the wave functions in the incompressible Schrödinger flow.