统计不相关最佳鉴别矢量集的本质研究

对统计不相关最佳鉴别矢量集的本质进行研究 ,在基于总体散布矩阵特征分解的基础上 ,构造了一种白化变换 ,使得变换后的样本空间中的总体散布矩阵为单位矩阵 ,这样使得传统的最佳鉴别矢量集算法得到的均是具有统计不相关的最佳鉴别矢量集 ,从而揭示了统计不相关最佳鉴别变换的本质———白化变换加普通的线性鉴别变换。该方法的最大优点在于所获得的最优鉴别矢量同时具有正交性和统计不相关性。该方法对代数特征抽取具有普遍适用性。用ORL人脸数据库的数值实验 ,验证了该方法的有效性

层状反倾向结构岩质边坡变形机理及稳定性研究

Toppling is a major failure model in anti-dip layered rock slopes. Because of the limited by testing means and experimental apparatus, present research on the deformation mechanism and stability analysis are mainly focus on the 2-Dimensional deformation, and the research really based on 3-Dimension is still limited. Therefore, based on the present research station, the article rely on the important hydroelectric project of typical anti-dip layered rock slopes -- The left bank slope of Long-tan hydropower-station in Guang Xi, China, and focused on the influencing factors, deformation mechanism and stability analysis of anti-dip layered rock slopes, three problems as follows are researched in this paper. (1) Deformation influencing factor analysis on ant-dip layered rock slopes Three influencing factors are included: geological factor, engineering factor and environmental factor. It is concluded that the toppling deformation of anti-dip layered rock slopes are more sensitive to geological and engineering factors, but less sensitive to environmental factor. In addition, the sensitivity of various factors to the rock toppling deformation is also arranged sequentially as follows: construction, gravitation, rainfall (underground water) and rock structure intensity, etc. (2) 3D deformation study on the anti-dip layered toppling rock slopes Used 3D Distinct Element Method (3DEC) analyzed the 3D deformation characteristic of anti-dip layered rock slops. It can be seen that the toppling characteristics are obvious when the inter-angle between slope direction and layer striking direction is under 20o, when the inter-angle is over 20o and equal or less than 40o，the toppling deformation characteristics decrease sharply with increase of inter-angle, when the inter-angle is over 40o , the slope deformation is not controlled by joints but influenced by other failure mode. Therefore, in order to quantify the toppling characteristics, a differential value of displacement vector angle between layered rock slope and block rock slope is proposed as a key index to distinguish failure model for anti-dip layered rock slopes, and it was applied to study the toppling of the rock slopes at Guangxi Long-tan hydropower station, China. The results indicated that the index was effective and instructive for analyzing the anti-dip layered rock slopes. (3) Stability analysis methods Because of the imperfection of some present slope analysis methods, based on slope failure mode and those three influencing factors, “slope stability entropy” method is defined in this paper, which makes good use of the sensitivity of relational matrix to influencing factors on slope stability and the qualification characteristics for information entropy to the irregularity of slope deformation. By this method, not only the randomness of geologic body on the base of dynamic analysis of slope failure mode is fully concerned, but also it makes the analysis time-saving and simple. Finally, the research findings were used to the engineering example successfully, and rational conclusion has been obtained.

粘弹性介质中的地震波传播与波形反演研究

The real earth is far away from an ideal elastic ball. The movement of structures or fluid and scattering of thin-layer would inevitably affect seismic wave propagation, which is demonstrated mainly as energy nongeometrical attenuation. Today, most of theoretical researches and applications take the assumption that all media studied are fully elastic. Ignoring the viscoelastic property would, in some circumstances, lead to amplitude and phase distortion, which will indirectly affect extraction of traveltime and waveform we use in imaging and inversion. In order to investigate the response of seismic wave propagation and improve the imaging and inversion quality in complex media, we need not only consider into attenuation of the real media but also implement it by means of efficient numerical methods and imaging techniques. As for numerical modeling, most widely used methods, such as finite difference, finite element and pseudospectral algorithms, have difficulty in dealing with problem of simultaneously improving accuracy and efficiency in computation. To partially overcome this difficulty, this paper devises a matrix differentiator method and an optimal convolutional differentiator method based on staggered-grid Fourier pseudospectral differentiation, and a staggered-grid optimal Shannon singular kernel convolutional differentiator by function distribution theory, which then are used to study seismic wave propagation in viscoelastic media. Results through comparisons and accuracy analysis demonstrate that optimal convolutional differentiator methods can solve well the incompatibility between accuracy and efficiency, and are almost twice more accurate than the same-length finite difference. They can efficiently reduce dispersion and provide high-precision waveform data. On the basis of frequency-domain wavefield modeling, we discuss how to directly solve linear equations and point out that when compared to the time-domain methods, frequency-domain methods would be more convenient to handle the multi-source problem and be much easier to incorporate medium attenuation. We also prove the equivalence of the time- and frequency-domain methods by using numerical tests when assumptions with non-relaxation modulus and quality factor are made, and analyze the reason that causes waveform difference. In frequency-domain waveform inversion, experiments have been conducted with transmission, crosshole and reflection data. By using the relation between media scales and characteristic frequencies, we analyze the capacity of the frequency-domain sequential inversion method in anti-noising and dealing with non-uniqueness of nonlinear optimization. In crosshole experiments, we find the main sources of inversion error and figure out how incorrect quality factor would affect inverted results. When dealing with surface reflection data, several frequencies have been chosen with optimal frequency selection strategy, with which we use to carry out sequential and simultaneous inversions to verify how important low frequency data are to the inverted results and the functionality of simultaneous inversion in anti-noising. Finally, I come with some conclusions about the whole work I have done in this dissertation and discuss detailly the existing and would-be problems in it. I also point out the possible directions and theories we should go and deepen, which, to some extent, would provide a helpful reference to researchers who are interested in seismic wave propagation and imaging in complex media.

大别山岩石圈有效弹性厚度初步研究

As powerful tools to study the lithosphere dynamics, the effective elastic thickness (Te) as well as the envelope of yielding stress of lithosphere have been attracted great attention of geoscientists in the past thirty years. The oceanic lithosphere, contrary to the continental lithosphere, has more fruits for its simple structures and evolution process. In continent, the lithosphere commonly is complex and variable in the rheological, thermal structures, and has a complicated history. Therefore, the application of the effective elastic thickness in continent is still a subject to learn in a long time. Te, with the definition of the thickness of an elastic plate in theory flexured by the equal benging of the real stress in the lithosphere plate (Turcotte, 1982), marks the depth of transition between elastic and fluid behaviors of rocks subjected to stress exceeding 100 MPa over the geological timescales (McNutt, 1990). There are three methods often adapted: admittance or isostatic response function, coherence and forwarding. In principle, the models of Te consist of thermal-rheological, non-linear Maxwell, non-linear work hardening and rheological layered models. There is a tentative knowledge of Te that it is affected by the following factors: crustal thickness, crust-mantle decoupling, plate bending, boundary conditions of plate (end forces and bending moments), stress state, sedimentary layer, faulting effect, variation in the mountain belts' strike, foreland basin, inheritance of tectonic evolution, convection of mantle, seismic depth and lithosphere strength. In this thesis, the author introduces the geological sketch of the Dabie collisional orogenic belt and the Hefei Basin. The Dabie Mts. is famous for the ultra-high pressure metamorphism. The crustal materials subducted down to the depth of at least 100 km and exhumed. So that the front subjects arise such as the deeply subduction of continent, and the post-collisional crust-mantle interaction. In a geological journey at June of 1999, the author found the rarely variolitic basaltic andesite in the Dabie Mts. It occurs in Susong Group, near Zhifenghe Countryside, Susong County, Anhui Province. It is just to the south of the boundary between the high-grade Susong melange and the ultra-high grade South Dabie melange. It has a noticeable knobby or pitted appearance in the surface. The size of the varioles is about 1-4 mm. In hand-specimen and under microscope, there are distinct contacts between the varioles and the matrice. The mineralogy of the varioles is primarily radiate plagioclase, with little pyroxene, hornblende and quartz. The pyroxene, hornblende and quartz are in the interstices between plagioclase. The matrix is consisted of glass, and micro-crystals of chlorite, epidote and zoisite. It is clearly subjected and extensive alteration. The andesite has an uncommon chemical composition. The SiO_2 content is about 56.8%, TiO_2 = 0.9%, MgO = 6.4%, (Fe_2O_3)_(Total) = 6.7% ～ 7.6%, 100 Mg/(Mg+Fe) = 64.1 ～ 66.2. Mg# is significantly high. The andesite has higher abundances of large-lithophile trace elements (e.g. K, Ba, Sr, LREE), e.g. La/Nd = 5.56-6.07, low abundances of high-strength-field elements (HFSE, e.g. Ta, Nb, P, Ti), particularly Ta and Nb strongly depleted. These are consistent with the characteristics of subducted-related magmas (Pearcce, 1982; Sun and McDonaugh, 1989). In the spider diagram of trace elements, from Ce to right hand, the abundances of elements decrease quickly, showing a characteristic of the continental margins (Pearce, 1982). There has a strongly enrichment of light-rare-earth elements, with a significant diffraction of REEs (the mean value of (La/Yb)_N is 32.84). No Eu anomaly, but there are anomaly high (La/Yb)_N = 28.63-36.74, (La/Y)_N = 70.33 - 82.84. The elements Y and Yb depleted greatly, Y < 20 ppm, Y_N = 2.74-2.84, Yb_N = 2.18 - 2.35. From the La-(La/Sm) diagram, the andesite is derived from partial melting. But the epsilone value of Nd is -18.7 ～ -19.2, so that the material source may be the mantle materials affected by the crustal materials. The Nd model age is 1.9 Ga indicating that the basaltic andesite was resulted from the post-collisional crust-mantle interaction between the subducted Yangze carton and the mantle of Sino-Korea carton. To obtain the Te of the lithosphere beneath the Dabie Mts. and the Hefei Basin, the author applies the coherence method in this thesis. The author makes two topography-gravity profiles (profiles 7~(th) and 9~(th)) across the Dabie Mts. and the Hefei Basin, and calculates the auto-coherence, across coherence, power spectrum, across power spectrum of the topography and gravity of the two profiles. From the relationships between the coherence and the wave-number of profiles. From the relationships between the coherence and the wave-number of profiles 7~(th) and 9~(th), it is obtained that the characteristic wavelengths respectively are 157 km and 126 km. Consequently the values of effective elastic thickness are 6.5 km and 4.8 km, respectively. However, the Te values merely are the minimum value of the lithosphere because the coherencemethod in a relative small region will generate a systemic underestimation. Why there is a so low Te value? In order to check the strength of the lithosphere beneath the Dabie Mts., the authore tries to outline the yielding-stress envelope of the lithosphere. It is suggested that the elastic layers in the crust and upper mantle are 18 km and 35 km, respectively. Since there exist a low viscosity layer about 3-5 km thickness, so it is reasonable that the decoupling between the crust and mantle occurred. So the effective thickness of the lithosphere can be estimated from the two elastic layers. Te is about 34 km. This is the maximum strength of the lithosphere. We can make an approximately estimation about the strength of the lithosphere beneath the Dabie Mts.: Te is about 20-30 km. The author believes that the following factors should be responsible for the low Te value: (1) the Dabie Mts. has elevated strongly since K_3-J_1. The north part of the Dabie Mts. elevates faster than the south part today; (2) there occur large active striking faults in this area. And in the east, the huge Tan-Lu striking fault anyway tends to decrease the lithosphere strength; (3) the lithosphere beneath the Dabie Mts. is heter-homogeneous in spatio-temporal; (4) the study area just locates in the adjacent region between the eastern China where the lithosphere thickness is significantly reduced and the normal western China. These factors will decrease the lithosphere strength.

Carbon nanotubes as adsorbent of solid-phase extraction and matrix for laser desorption/ionization mass spectrometry

A method with carbon nanotubes functioning both as the adsorbent of solid-phase extraction (SPE) and the matrix for matrix assisted laser desorption/ ionization mass spectrometry (MALDI-MS) to analyze small molecules in solution has been developed. In this method, 10 muL suspensions of carbon nanotubes in 50% (vol/vol) methanol were added to the sample solution to extract analytes onto surface of carbon nanotubes because of their dramatic hydrophobicity. Carbon nanotubes in solution are deposited onto the bottom of tube with centrifugation. After removing the supernatant fluid, carbon nanotubes are suspended again with dispersant and pipetted directly onto the sample target of the MALDI-MS to perform a mass spectrometric analysis. It was demonstrated by analysis of a variety of small molecules that the resolution of peaks and the efficiency of desorption/ ionization on the carbon nanotubes are better than those on the activated carbon. It is found that with the addition of glycerol and sucrose to the dispersant, the intensity, the ratio of signal to noise (S/N), and the resolution of peaks for analytes by mass spectrometry increased greatly. Compared with the previously reported method by depositing sample solution onto thin layer of carbon nanotubes, it is observed that the detection limit for analytes can be enhanced about 10 to 100 times due to solid-phase extraction of analytes in solution by carbon nanotubes. An acceptable result of simultaneously quantitative analysis of three analytes in solution has been achieved. The application in determining drugs spiked into urine has also been realized. (C) 2004 American Society for Mass Spectrometry.

Using oxidized carbon nanotubes as matrix for analysis of small molecules by MALDI-TOF MS

Oxidized carbon nanotubes are tested as a matrix for analysis of small molecules by matrix assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF-MS). Compared with nonoxidized carbon nanotubes, oxidized carbon nanotubes facilitate sample preparation because of their higher solubility in water. The matrix layer of oxidized carbon nanotubes is much more homogeneous and compact than that of nonoxidized carbon nanotubes. The efficiency of desorption/ionization for analytes and the reproducibility of peak intensities within and between sample spots are greatly enhanced on the surface of oxidized carbon nanotubes. The advantage of the oxidized carbon nanotubes in comparison with alpha-cyano-4-hydroxycinnamic acid (CCA) and carbon nanotubes is demonstrated by MALDI-TOF-MS analysis of an amino acid mixture. The matrix is successfully used for analysis of synthetic hydroxypropyl P-cyclodextrin, suggesting a great potential for monitoring reactions and for product quality control. Reliable quantitative analysis of jatrorrhizine and palmatine with a wide linear range (1-100 ng/mL) and good reproducibility of relative peak areas (RSD less than 10 %) is achieved using this matrix. Concentrations of jatrorrhizine (8.65 mg/mL) and palmatine (10.4 mg/mL) in an extract of Coptis chinensis Franch are determined simultaneously using the matrix and a standard addition method. (c) 2005 American Society for Mass Spectrometry.

Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry with a matrix of carbon nanotubes for the analysis of low-mass compounds in environmental samples

The use of matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) for environmental analysis has been mainly focused on qualitative analysis of high-mass molecules, such as toxins, humic acid, and microorganisms. Herein,we describe a novel MALDI-TOF-MS method with a matrix of oxidized carbon nanotubes for analysis of low-mass compounds in environmental samples. A number of chemicals in the environment were qualitatively analyzed by the present method, and it was found that most of them, especially the highly polar chemicals, were measurable with high sensitivity. With the intrinsic ability to measure high-mass chemicals, this method can compensate for the current shortage of methods for environmental analysis for the measurement of highly polar or high-mass chemicals. For sample analysis, arsenic speciation in Chinese traditional medicines was qualified and diphenylolpropane in water samples was quantified. With the relatively high tolerance of the method to interfering molecules, a simple pretreatment or even no pretreatment could be employed before MS detection. Furthermore, this method can be employed in a high-throughput format.

Separation and detection of compounds in Honeysuckle by integration of ion-exchange chromatography fractionation with reversed-phase liquid chromatography-atmospheric pressure chemical ionization mass spectrometer and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry analysis

A hyphenated method for the isolation and identification of components in a traditional Chinese medicine of Honeysuckle was developed. Ion-exchange chromatography (IEC) was chosen for the fractionation of Honeysuckle extract, and then followed by concentration of all the fractions with rotary vacuum evaporator. Each of the enriched fractions was then further analyzed by reversed-phase liquid chromatography-atmospheric pressure chemical ionization mass spectrometer (RPLC-APCI/MS) and matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF/MS) with matrix of oxidized carbon nanotubes, respectively. It can be noted totally more than 117 components were detected by UV detector, APCI/MS and MALDI-TOF/MS in Honeysuckle extract except the, 145 components identified by MALDI-TOF/MS alone with this integrated approach, and 7 of them were preliminary identified according to their UV spectra and mass spectra performed by APCI/MS and MALDI-TOF/MS, respectively. The obtained analytical results not only indicated the approach of integration IEC fractionation with RPLC-APCI/MS and MALDI-TOF/MS is capable of analyzing complex samples, but also exhibited the potential power of the mass spectrometer in detection of low-mass compounds, such as traditional Chinese medicines (TCMs) and complex biological samples. (c) 2005 Elsevier B.V. All rights reserved.

Development and application of a sol-gel immunosorbent-based method for the determination of isoproturon in surface water

An immunosorbent was fabricated by encapsulation Of monoclonal anti-isoproturon antibodies in sol-gel matrix. The immunosorbent-based loading, rinsing and eluting processes were optimized. Based on these optimizations, the sol-gel immunosorbent (SG-IS) selectively extracted isoproturon from an artificial mixture of 68 pesticides. In addition to this high selectivity, the SG-IS proved to be reusable. The SG-IS was combined with liquid chromatography-tandem mass spectrometry (LC-MS-MS) to determine isoproturon in surface water, and the linear range was up to 2.2 mu g/l with correlation coefficient higher than 0.99 and relative standard deviation (RSD) lower than 5% (n = 8). The limit of quantitation (LOQ) for 25-ml surface water sample was 5 ng/l. (c) 2006 Elsevier B.V. All rights reserved.