利用AMSR-E被动微波数据反演地表温度的神经网络算法

结合对地观测卫星AQUA多传感器/多分辨率的特点,研究了利用AMSR-E被动微波数据反演地表温度的神经网络算法.MODIS地表温度(LST)产品被作为地表温度实测数据,对应的平均温度被用作对应AMSR-E像元的实际地表温度,从而克服由于AMSR-E像元尺度太大和云的影响而难以获得地表实测数据的难点.反演结果分析表明,利用神经网络能够精确地由AMSR-E数据反演地表温度.当使用5个频率10个通道反演时,反演精度最高,说明使用更多的通道能更好地消除土壤水分、粗糙度、大气和其它因素的影响.相对于MODIS温度产品,用此算法反演的平均误差约低于2K.     

基于神经网络方法的高光谱遥感浅海水深反演

利用人工神经网络方法进行了高光谱遥感反演浅海水深的初步研究.在产生模拟数据时,为保证模拟数据的合理性,引入了根据水体和海底特性来划分光学浅水和光学深水的方法,并初步研究了利用光谱徽分技术进行光学浅水和光学深水区分的有效性.在人工神经网络建模过程中,采用主成分分析的方法对网络的输入数据进行预处理,显著提高了网络的学习速度.建立的人工神经网络模型和基于非线性最优化方法的反演算法与实测数据的反演结果相比较,人工神经网络模型的反演精度明显高于非线性最优化反演算法.     

Evaluation of the depth resolutions of Auger electron spectroscopic, X-ray photoelectron spectroscopic and time-of-flight secondary-ion mass spectrometric sputter depth profiling techniques

Concentration-depth profiles of sputter-deposited Si/Al multilayered specimens were determined by model fitting to measured data obtained by depth profiling, using Auger electron spectroscopy (AES), X-ray photoelectron spectroscopy (XPS) and time-of-flight secondary-ion mass spectrometry (TOF-SIMS). The model used for calculation of the concentration-depth profile accounts for the broadening (“smearing”) upon experimental depth profiling owing to the effects of atomic mixing, preferential sputtering, surface roughness and information depth of either the Auger electrons (for AES depth profiling) or the photoelectrons (for XPS depth profiling) or the secondary ions (for SIMS depth profiling). The depth resolution for each technique was derived directly from the values determined for the fitting parameters in the model.     

A simplified method for predicting the stress concentration in notches from experimental stress data

This paper describes a simple procedure to obtain very reliable estimates of the stress concentration at the root of a notch using only data from points beneath the notch root along its axis of symmetry. The advantages of the method are (i) data from only a few points are needed, (ii) the necessary computations require only a hand held calculator, (iii) inaccurate data becomes quite apparent to the analyst, and (iv) random experimental errors are largely compensated.     

A new method for determining the sharpness of InGaAs/GaAs heterojunctions by auger depth profiling

It is established for the first time that the phenomenon of ion-stimulated surface segregation can be used to increase the depth resolution of Auger profiling during analysis of the In_x Ga_{1 − x} As/GaAs heterostructures. It is demonstrated that, by varying the energy of the sputtering Ar⁺ ion beam from 1 to 0.5 keV in the region of the GaAs/InGaAs heterojunction, the junction sharpness can be estimated at a resolution on the order of 0.5 nm determined by a difference in the projected range of Ar⁺ ions and independent of the escape depth of the Auger electrons.     

Optoelectronic neural system for processing the output data from a fiber-optic measuring network

The results of an investigation of an optoelectronic neural system for processing the output data from a distributed fiber-optic tomographic-type measuring network are reported. The processing system is based on the operation principle of an optical perceptron in which the interneuron coupling matrix is implemented using a collection of amplitude holograms recorded on a disk-shaped holographic carrier. It is shown experimentally that this system makes it possible to reconstruct to within 20% the spatial distribution functions of the physical quantity under investigation. Pis’ma Zh. Tekh. Fiz. 25, 65–70 (March 26, 1999)     

Modular artificial neural network for prediction of petrophysicalproperties from well log data

An application of Kohonen's self-organizing map (SOM), learning-vector quantization (LVQ) algorithms, and commonly used backpropagation neural network (BPNN) to predict petrophysical properties obtained from well-log data are presented. A modular, artificial neural network (ANN) comprising a complex network made up from a number of subnetworks is introduced. In this approach, the SOM algorithm is applied first to classify the well-log data into a predefined number of classes, This gives an indication of the lithology in the well. The classes obtained from SOM are then appended back to the training input logs for the training of supervised LVQ. After training, LVQ can be used to classify any unknown input logs. A set of BPNN that corresponds to different classes is then trained. Once the network is trained, it is then used as the classification and prediction model for subsequent input data. Results obtained from example studies using the proposed method have shown to be fast and accurate as compared to a single BPNN network     

A method for determination of ion distribution within electrosprayed droplets

A simplified method for determination of ion distribution within electrospray ionization droplets has been developed. The method is based on an electrospray ionization mass spectrometry equilibrium partitioning model recently developed by Enke (Enke, C. G. Anal. Chem. 1997, 69, 4885-4893). This "simple" method required only two samples to be analyzed at each solution composition compared to the method previously reported. Furthermore, as the same m/z value is monitored in both experiments, possible effects of the mass-dependent ion transmission in the quadrupole were eliminated. Tetra-alkylammonium compounds with variable hydrophobicity were used as model compounds and the effect of methanol-water composition in the electrosprayed solution was studied. It was found, as expected, that the signal optimized at a high content of methanol in the electrosprayed solution. The distribution coefficient for analyte ions between the droplet surface and bulk solution maximize, however, at a lower content of methanol in the electrosprayed solution.     

EFSA—a new evaluation method of the depth resolution in depth profiling of multilayer structures

A new evaluation method called EFSA (error function superposition approximation) is proposed, to obtain depth resolution functions on any multilayer structure in agreement with the conventional absolute depth resolution. The quantitative evaluation of Δz measured on multilayer samples helps in clearing up the relations of different interface broadening effects and makes the evaluation independent of the measuring system. It is shown that this method can be successfully applied to both SIMS and AES depth profile measurements and that it has proved to be useful in comparing results obtained in different laboratories.     

A neural network approach for the prediction of the refractive index based on experimental data

This article presents a systematic approach for correlating the refractive index of different material kinds and forms with experimentally measured inputs like wavelength, temperature, and concentration. The correlation is accomplished using neural network models, which can deal effectively with the nonlinear nature of the problem without requiring a predefined form of equation, while taking into account all the parameters affecting the refractive index. The proposed methodology employs the powerful radial basis function network architecture and the neural network training procedure is accomplished using an innovative algorithm, which provides results with increased prediction accuracy. The methodology is applied to two cases, involving the estimation of the refractive index of semiconductor material crystals and an ethanol–water mixture and the results show that the refractive index predictions are accurate approximately to the same number of decimal places as the real measurements. Comparisons with other neural network training methods, but also with empirical forms like the Sellmeier equation, highlight the superiority of the proposed approach.     

A hybrid artificial neural network method with uniform design for structural optimization

This paper presents a new hybrid artificial neural network (ANN) method for structural optimization. The method involves the selection of training datasets for establishing an ANN model by uniform design method, approximation of the objective or constraint functions by the trained ANN model and yields solutions of structural optimization problems using the sequential quadratic programming method (SQP). In the proposed method, the use of the uniform design method can improve the quality of the selected training datasets, leading to a better performance of the ANN model. As a result, the ANN dramatically reduces the number of required trained datasets, and shows a good ability to approximate the objective or constraint functions and then provides an accurate estimation of the optimum solution. It is shown through three numerical examples that the proposed method provides accurate and computationally efficient estimates of the solutions of structural optimization problems.     

Computationally efficient method for retrieving aerosol optical depth from ATSR-2 and AATSR data

We present a robust and computationally efficient method for retrieving aerosol optical depth (AOD) from top-of-atmosphere ATSR-2 (Along-Track Scanning Radiometer) and AATSR (Advanced ATSR) reflectance data that is formulated to allow retrieval of the AOD from the 11 year archive of (A)ATSR data on the global scale. The approach uses a physical model of light scattering that requires no a priori information on the land surface. Computational efficiency is achieved by using precalculated lookup tables (LUTs) for the numerical inversion of a radiative-transfer model of the atmosphere. Estimates of AOD retrieved by the LUT approach are tested on AATSR data for a range of global land surfaces and are shown to be highly correlated with sunphotometer measurements of the AOD at 550 nm. (Pearson's correlation coefficient r(2) is 0.71.).     

A new alternative to secondary CsM+ ions for depth profiling of multilayer metal structures by secondary ion mass spectrometry

A comparison of the results of depth profiling of multilayer metal structures by secondary ion mass spectrometry with the use of various types of the registered secondary ions has been conducted. For the first time, it has been demonstrated that, to increase depth resolution, two variants can be used in addition to the known secondary ions CsM⁺ (M = La, Pd, Mo): M⁺ at sputtering by cesium ions and OM^? at sputtering by oxygen ions. For the Mo/Si multilayer structures, the use of the elementary secondary ions Mo⁺ and Si⁺ at sputtering by Cs ions and probing by the cluster ions provides the best depth resolution.     

An exploratory study of a neural network approach for reliability data analysis

The results of this paper show that neural networks could be a very promising tool for reliability data analysis. Identifying the underlying distribution of a set of failure data and estimating its distribution parameters are necessary in reliability engineering studies. In general, either a chi-square or a non-parametric goodness-of-fit test is used in the distribution identification process which includes the pattern interpretation of the failure data histograms. However, those procedures can guarantee neither an accurate distribution identification nor a robust parameter estimation when small data samples are available. Basically, the graphical approach of distribution fitting is a pattern recognition problem and parameter estimation is a classification problem where neural networks have been proved to be a suitable tool. This paper presents an exploratory study of a neural network approach, validated by simulated experiments, for analysing small-sample reliability data. A counter-propagation network is used in classifying normal, uniform, exponential and Weibull distributions. A back-propagation network is used in the parameter estimation of a two-parameter Weibull distribution.     

A combination of rough-based feature selection and RBF neural network for classification using gene expression data

This paper presents a novel rough-based feature selection method for gene expression data analysis. It can find the relevant features without requiring the number of clusters to be known a priori and identify the centers that approximate to the correct ones. In this paper, we attempt to introduce a prediction scheme that combines the rough-based feature selection method with radial basis function neural network. For further consider the effect of different feature selection methods and classifiers on this prediction process, we use the NaIve Bayes and linear support vector machine as classifiers, and compare the performance with other feature selection methods, including information gain and principle component analysis. We demonstrate the performance by several published datasets and the results show that our proposed method can achieve high classification accuracy rate.     

基于爆破振动的岩质边坡损伤神经网络预测

岩石高边坡的爆破开挖会对保留岩体造成损伤,岩体损伤过大可能导致边坡失稳,需严格控制并准确确定开挖损伤深度,因此,提出一种快速精确的损伤深度预测方法。以白鹤滩水电站左岸834.0~770.0 m高程坝肩槽边坡爆破开挖为背景,利用六个开挖梯段的多高程、多爆心距爆破振动监测及损伤深度声波检测的数据,建立基于振动峰值的爆破损伤深度BP神经网络预测模型,对高边坡爆破损伤深度进行实时预测。该方法利用不同部位及不同爆心处的质点峰值振动峰值作为主回归变量,同时还考虑最大单响药量和岩体强度的影响。结果表明,当开挖区域坡体岩性相似且无长大软弱结构面发育时,运用神经网络模型及多高程实测爆破振动预测本梯段爆破损伤深度的方法简便可行,预测精度可满足实际工程需求。作为传统爆破损伤声波检测的补充,可大大减轻现场声波测试工作量。     

基于爆破振动的岩质边坡损伤神经网络预测

岩石高边坡的爆破开挖会对保留岩体造成损伤,岩体损伤过大可能导致边坡失稳,需严格控制并准确确定开挖损伤深度,因此,提出一种快速精确的损伤深度预测方法。以白鹤滩水电站左岸834.0⁷70.0 m高程坝肩槽边坡爆破开挖为背景,利用六个开挖梯段的多高程、多爆心距爆破振动监测及损伤深度声波检测的数据,建立基于振动峰值的爆破损伤深度BP神经网络预测模型,对高边坡爆破损伤深度进行实时预测。该方法利用不同部位及不同爆心处的质点峰值振动峰值作为主回归变量,同时还考虑最大单响药量和岩体强度的影响。结果表明,当开挖区域坡体岩性相似且无长大软弱结构面发育时,运用神经网络模型及多高程实测爆破振动预测本梯段爆破损伤深度的方法简便可行,预测精度可满足实际工程需求。作为传统爆破损伤声波检测的补充,可大大减轻现场声波测试工作量。     

Determination of the diameter distribution of single-wall carbon nanotubes from the Raman G-band using an artificial neural network

A novel, artificial neural network-based method is now available for obtaining the mean diameter of single wall carbon nanotube (SWCNT) samples from the diameter dispersive features of their Raman G-band. The method is demonstrated here for six different diameter SWCNT samples and 14 different excitation wavelengths. With an adequately large pool of standard nanotube samples, the suggested method is a useful complementary technique for SWCNT diameter analysis as it is capable of rapid diameter evaluation without prior knowledge of the relevant phonon dispersion relations.     

A method for determination of spatial distribution of gas concentration in supersonic jets outflowing into vacuum

An experimental method for determination of the spatial distribution of gas concentration in millimeter-sized supersonic jets outflowing into vacuum is proposed. This method is based on jet visualization by its illumination with a laser–plasma radiation source and processing of the obtained intensity distribution of the jet glow. The spatial distribution of the jet glow intensity for the visualized jet is measured. The gas concentration distribution in the jet is determined, and the results of experiment are compared with the calculated data obtained by numerical hydrodynamic simulation.     

Modeling of material behavior data in a functional form suitable for neural network representation

Decision making through connectionist expert systems is getting adopted in various disciplines. For such systems the subject knowledge has to be represented by neural networks. A simple and fast method is developed for modeling of material behavior data in a functional form suitable for neural network representation. Complete algorithm for the method is given. The algorithm is used for modeling of material behavior data for several examples, taken from handbooks, etc. The error of approximation is very small in all the cases considered. The algorithm deals with cases where the property to be modeled is a function of one variable or its variation is measured with change of one parameter, keeping other parameters constant. The algorithm can also be used as a curve fitting technique, preserving the shape as desired by subject expert, even in presence of errors in the measurement data.