基于BP神经网络的三维激光测高数据拟合方法

针对三维激光测高拟合问题，本文采用BP神经网络进行拟合，该方法可以有效地消除或减弱数据中的系统误差影响，提高计算精度。实际算例证明，在数据无粗差情况下，BP神经网络的内符合与外符合精度优于传统的二次曲面拟合的精度。     

矢量数据动态存贮方法的研究

本文通过对三种矢量数据动态存贮方法的分析比较，并通过实践，找到一种安全可靠并且比较实用的数据动态存贮方式。     

基于自适应Kalman滤波的BP神经网络及其在高程拟合中的应用

针对传统BP神经网络容易发散、泛化能力差等问题，采用自适应Kalman滤波方法训练神经网络的连接权。与Kalman滤波训练连接权和传统的BP算法相比，该方法提高了BP神经网络计算精度，增强了泛化能力。实测数据的计算结果证明了该方法的可行性与有效性。     

Modelling the 10-image lensed system B1933+503

A gravitational lens model is presented for the newly discovered 10-image system B1933+503. The underlying object, revealed by modelling, is a triple radio source on the scale of a couple of hundred mas that is well-aligned along the line of sight with a foreground and somewhat flattened lensing galaxy, the orientation and location of which match those of an observed galaxy, known to be at a redshift of 0.755. Uncertainties in the modelling are obtained by a Monte Carlo exercise. Observational tests of the lens model are proposed, and the time delays between various pairs of images are determined, as the core of the source is known to be significantly variable. Future observations of the lens hold the key to using B1933+503 to constrain Hubble's constant. Despite the absence of a source redshift, the utility of the system as a probe of the structure of the lens galaxy is unparalleled as it provides a surfeit of easily identifiable constraints for modelling the system.     

The non-linear redshift-space power spectrum of galaxies

We study the power spectrum of galaxies in redshift space, with third-order perturbation theory to include corrections that are absent in linear theory. We assume a local bias for the galaxies: i.e., the galaxy density is sampled from some local function of the underlying mass distribution. We find that the effect of the non-linear bias in real space is to introduce two new features: first, there is a contribution to the power which is constant with wavenumber, whose nature we reveal as essentially a shot-noise term. In principle this contribution can mask the primordial power spectrum, and could limit the accuracy with which the latter might be measured on very large scales. Secondly, the effect of second- and third-order bias is to modify the effective bias (defined as the square root of the ratio of galaxy power spectrum to matter power spectrum). The effective bias is almost scale-independent over a wide range of scales. These general conclusions also hold in redshift space. In addition, we have investigated the distortion of the power spectrum by peculiar velocities, which may be used to constrain the density of the Universe. We look at the quadrupole-to-monopole ratio, and find that higher order terms can mimic linear theory bias, but the bias implied is neither the linear bias, nor the effective bias referred to above. We test the theory with biased N -body simulations, and find excellent agreement in both real and redshift space, providing the local biasing is applied on a scale whose fractional rms density fluctuations are < 0.5.     

三峡库区农业生态经济系统的预警分析

从系统与环境相统一的角度,研究了农业生态经济系统的模型、特征、预警分析的意义和内容;以三峡库区为例,从系统序化的观点,确定了这个系统的社会、经济、环境功能和状态的主要指标的警戒线;并与这些指标的现状、过去与未来趋势值进行对比;分别对现状预警、趋势预警和突变预警进行了评价;最后作了简易的对策探讨．     

用附有条件的参数平差法求解秩亏网新探

通过引进典则参数,以较简捷的方式证明了秩亏网平差应满足的条件方程,从而用附有条件的平差方法求解秩亏网平差参数的最小二乘极小范数解。     

经济信息管理系统分析与研究

根据经济信息管理系统的组织特征与建立原则 ,提出经济信息管理系统的结构模式与建立步骤。