基于DFA的我国股票市场标度特性研究

将消除趋势波动分析法(DFA)应用于我国沪深两市不同时间标度收盘指数的对数收益率序列,全面分析了我国股市的标度特性.发现标度指数随着考察时间的长短,即数据个数的不同而不同,但从长期来看,两市波动都存在持久性特征,而且收益率的绝对值序列和平方值序列的持久性更明显;在相同时间范围内,不同时间标度序列的标度指数相差不大,存在标度不变性;小标度时间序列存在多重分形特征.这些结果说明我国股票市场存在复杂的非线性动力学特性.     

基于数据列变换的自回归预测方法的改善

证明了利用反双曲正弦函数变换能提高数据列的光滑程度,获得结论:设｛x_(k)｝为递增数据列,x_(1)>0,y(k)=ln(x_(k) (x_(k)~2 1)~(1/2)),则数据列｛y_(k)｝比数据列｛x_(k)｝光滑.给出了改善的自回归预测方法,并且举例加以论证.     

基于特征加权的神经网络集成及其应用

泛化能力是机器学习关注的基本问题之一.特征加权是特征选择的更一般情况,它能更加细致地区分特征对结果影响的程度,往往能够获得比特征选择更好的或者至少相等的性能,已经成为普遍的提高学习器的泛化能力的方法之一.该文提出一种基于特征加权的神经网络集成方法FWEART,该方法通过自适应遗传算法的优胜劣汰机制为输入属性确定了特征权值,提高了集成中各个体Category ART网络的精度和差异度,从而提高了神经网络集成的泛化能力.在UCI标准数据集上验证了有效性后,FWEART被应用在地震序列类型预报上,取得了较好的预报效果.     

二维楔形体入水砰击仿真研究

基于数值仿真的方法对二维刚性楔形体的入水砰击问题进行了研究.建立了包含气水流场的二维有限元模型,考虑空气、重力等因素在砰击过程中的影响,研究了流场的自由液面、压力和速度等物理量的变化情况.对不同入水角的楔形体入水的砰击压力峰值系数进行了计算,分析压力峰值的变化规律.结果表明,楔形体入水砰击压力峰值系数不仅与入水角度有关,而且与入水速度也有很大的关系.     

基于可解释机器学习的水平井产能预测方法

准确预测致密气藏分段压裂水平井产能是压裂效果评价和优化设计的关键环节。现有的产能预测方法,引入了过多的假设和简化,很难全面反映致密储层流体多尺度的运移机理和复杂物理过程,导致产能预测误差较大。提出一种基于机器学习的致密气藏分段压裂水平井产能预测方法,该方法综合利用已收集的地质、压裂水平井产能及钻完井等多类型数据,通过机器学习算法直接挖掘数据内部规律,建立产能预测模型。此外,为解决常规机器学习模型的“黑盒子”问题,还利用SHAP(SHapley Additive exPlanations)方法对建立的机器学习模型进行全局和局部解释,分析影响产能的主要因素,增加了模型的可信性和透明度。以苏里格气田苏东示范区为例,验证了该方法的有效性和实用性。与油气藏数值方法相比,该方法不仅提高了产能预测的精度,而且缩短了建模周期,加快了计算速度。     

Water Status and Predictive Models of Moisture Content during Drying of Soybean Dregs Based on LF-NMR

To explore the drying characteristics of soybean dregs and a nondestructive moisture content test method, in this study, soybean dregs were dried with hot air (80 °C), the moisture content was measured using the drying method, water status was analyzed using low-field nuclear magnetic resonance (LF-NMR) and the moisture content prediction models were built and validated. The results revealed that the moisture contents of the soybean dregs were 0.57 and 0.01 g/g(w.b.), respectively, after drying for 5 and 7 h. The effective moisture diffusivity increased with the decrease in moisture content; it ranged from 5.27 × 10⁻⁹ to 6.96 × 10⁻⁸ m²·s⁻¹. Soybean dregs contained bound water (T₂₁), immobilized water (T₂₂) and free water (T₂₃ and T₂₃’). With the proceeding of drying, all of the relaxation peaks shifted left until a new peak (T₂₃’) appeared; then, the structure of soybean dregs changed, and the relaxation peaks reformed, and the peak shifted left again. The peak area may predict the moisture content of soybean dregs, and the gray values of images predict the moisture contents mainly composed of free water or immobilized water. The results may provide a reference for drying of soybean dregs and a new moisture detection method.     

黑龙江省通信客户发展分析预测

通过运用灰色系统模型对黑龙江省通信公司近年来的客户发展进行预测,对预测结果进行营销分析,同时寻找如何增强并保持客户的能力,这将对通信企业未来的生产计划产生深远影响,同时也对我省电信增值服务,通信产品发展具有重要的参考价值.     

Exploring the Parallel G-Quadruplex Nucleic Acid World: A Spectroscopic and Computational Investigation on the Binding of the c-myc Oncogene NHE III1 Region by the Phytochemical Polydatin

Trans-polydatin (tPD), the 3-β-D-glucoside of the well-known nutraceutical trans-resveratrol, is a natural polyphenol with documented anti-cancer, anti-inflammatory, cardioprotective, and immunoregulatory effects. Considering the anticancer activity of tPD, in this work, we aimed to explore the binding properties of this natural compound with the G-quadruplex (G4) structure formed by the Pu22 [d(TGAGGGTGGGTAGGGTGGGTAA)] DNA sequence by exploiting CD spectroscopy and molecular docking simulations. Pu22 is a mutated and shorter analog of the G4-forming sequence known as Pu27 located in the promoter of the c-myc oncogene, whose overexpression triggers the metabolic changes responsible for cancer cells transformation. The binding of tPD with the parallel Pu22 G4 was confirmed by CD spectroscopy, which showed significant changes in the CD spectrum of the DNA and a slight thermal stabilization of the G4 structure. To gain a deeper insight into the structural features of the tPD-Pu22 complex, we performed an in silico molecular docking study, which indicated that the interaction of tPD with Pu22 G4 may involve partial end-stacking to the terminal G-quartet and H-bonding interactions between the sugar moiety of the ligand and deoxynucleotides not included in the G-tetrads. Finally, we compared the experimental CD profiles of Pu22 G4 with the corresponding theoretical output obtained using DichroCalc, a web-based server normally used for the prediction of proteins’ CD spectra starting from their “.pdb” file. The results indicated a good agreement between the predicted and the experimental CD spectra in terms of the spectral bands’ profile even if with a slight bathochromic shift in the positive band, suggesting the utility of this predictive tool for G4 DNA CD investigations.     

乌里雅斯太凹陷南洼槽低渗透油藏勘探研究方法及勘探效果

针对乌里雅斯太凹陷南洼槽储层物性差、非均质性强、油藏类型以岩性和地层油藏为主的低渗透油藏特点,以沉积储层为主要研究内容,通过层序地层学方法和储层预测新技术的应用,识别出多个不同时期、不同类型的砂体,圈定了砂体范围,预测了有利储层部位。在3年的勘探实践中,先后钻探了10余口探井,全部获得成功,勘探取得了重大突破。