基于随机森林的高寒草地地上生物量高光谱估算

草地地上生物量（Aboveground biomass,AGB）是衡量草地生产力的关键因素,准确测定草地AGB具有重要意义。高光谱因具有时效性强、不破坏草地等特点被广泛用于草地生理生态指标的测定。本研究提取和计算了海北试验站高寒草地冠层的原始光谱（Original spectrum,OR）反射率、一阶微分光谱（First derivative spectrum,FD）反射率、光谱位置面积参数（Spectral parameters of spectral position and area,PA）和植被指数（Vegetation indices,VI）4种不同类型的特征变量,使用连续投影算法（Successive projections algorithm,SPA）和递归特征消除算法（Recursive feature elimination,RFE）进行特征选择,采用随机森林算法（Random forest,RF）构建草地AGB估测模型。结果表明：在由4种特征变量分别构建的草地AGB估测模型中,基于VI的RF模型精度最高（测试集R²=0.70,RMSE=557.87 kg·ha^-1）,实测AGB与估测AGB的线性R²达到0.72;不同类型特征变量组合构建的草地AGB估测模型中,PA+VI组合的RF模型精度最高（R²=0.71,RMSE=548.97 kg·ha^-1）,实测AGB和估测AGB的线性R²达到0.73。     

基于模拟退火算法优化给水管线造价公式研究

给水管线造价公式的精确性对工程经济分析与给水管网优化的科学性和经济性有重大影响。通过五个地区的铸铁管综合单价数据,采用最小二乘法对各组数据进行多项式一至十次的拟合,确定了给水管线造价公式的最佳拟合次数为三次;进一步采用模拟退火算法对拟合的多项式参数进行优化计算,求解给水管线造价公式。结果表明:凭借模拟退火算法随机全局搜索模式以及不受函数性质影响的优势,能够克服传统算法难以求解多阶导数的困难,提高公式拟合精度。     

基于RAGA的PPC模型在黑土区玉米调亏灌溉评价中的应用

调亏灌溉作为一种行之有效的节水灌溉方式,对黑土区农业水资源的可持续高效利用具有重要意义。而调亏灌溉方案的优劣取决于它的综合效果,难以直观评价。以2013年在黑龙江省红星农场进行的玉米调亏灌溉技术效应的试验研究为基础,选取了8个评价指标,采用投影寻踪分类模型(PPC),利用基于实数编码的加速遗传算法(RAGA)优化其投影方向,根据投影函数值得到各评价指标对综合评价的贡献大小对不同调亏处理的各项指标进行排序。结果表明:苗期水分调亏程度为田间持水量60%的处理为最佳调亏灌溉方案,产量和水分利用效率比适宜调亏灌溉处理分别提高了2.11%和6.42%。研究结果可为黑土区玉米灌溉管理提供科学依据。     

考虑陀螺效应的动力总成悬置系统分析和优化

应用陀螺力学理论，对传统动力总成悬置系统的无阻尼动力学方程进行修正，引入陀螺矩阵，得到新的动力学方程。对新的动力总成悬置系统动力学模型进行计算分析，研究了陀螺矩阵对频域、时域响应的影响。使用遗传算法分别对修改前后的方程进行优化计算，并进行比较分析。结果表明，对于低转速发动机，曲轴转动的陀螺效应不明显，而对于高转速的发动机，陀螺效应不能忽略，应以考虑陀螺力矩的方程为基础进行设计分析。     

Wavelet Application in Algorithms of Data Mining

Wavelets have been widely applied in such research areas as image processing, computer vision, network management and data mining. Wavelets have many favorable properties, which are hierarchical and multiresolution decomposition structure, linear time and space complexity of the transformations. The paper presents wavelet application on the algorithms of data mining, in cluding as clustering, classification, distributed data mining, similarity search, approximate query processing, visualization and so on. Finally, the paper concludes by discussing the impact of wavelets on data mining research and outlining potential future research directions and applications.     

A MIMD PARALLEL ALGORITHM TO SOLVE TRIDIAGONAL TOEPLITZ LINEAR EQUATIONS

The rank one updating formula and the parallel eliminating algorithm for linear recurrence systems ane combined fo form a MTMD parallel algorithm fo solve tridiagonal Toeplitz Cinear equations. This algorithm has a simple strictire and requires onlty a few storages as well as in-terprocessor communications. There is no special demand on the number of processors in the parallel system. The speedup can come dose to p12 comparing with the LU decomposition method (p is the number of processors). It is worth pointing out that it is proposed algcmthn that considels and organizes the parallel arithmetic in the first order linear ricurrence systems with constant coefficients.     

A FINITE ELEMENT ALGORITHM FOR PARALLEL PROCESSING OF MULTI-FRONTS

In this paper,a new algorithm for solving large-scale finite element system is presented, which utilizes the architecture of parallel processing. Firstly, the structure is partitioned into sub-regions. Then multi-fronts are used to assemble and eliminate concurrently in every sub-region. Thus the condensed interface stiffness matrices and load vectors are obtained. By serially assembling and solving global equation to get interface displacements and backing to every sub-region, the inter-nodal displacements and element stresses can be solved. The results of computation show that it can not only increase the speed of computation,but also save the memory space of computer effectively. This algorithm is an effective method for solving large-scale structural finite element systems.     

NETWORK FEOW APPROACH TO RELIABILITY EVALUATION IN IARGE SCALE NETWORKS BY MIXING INCREASING ELOW WITH DECREASING FLOW

Based on the maximum flow principle about networks, a network flow approach to reliability evaluation in large scale netwerks ty mixing inereasing flow with decreasing flow is proposed in this paper.This approach gets the initial flow of each branch in a r.etwork associated with a state of the retwork by decreasing flows of some branches correlative to another state of the network.As a result, the number of flow compute required can be reduced enormously.The numerical example shows that the approach is powerful. The comparison of this approach with the minimal cutset one is also given in this paper.     

A Study on Human Simulation Intelligent Control Algorithm to a Category of Objects

Based on the model of Human Simulation Intelligent Controller (HSIC) presented briefly, a detailed algorithm was described for the servo system control by using the method of HSIC. Simulation results obtained proves the good performance of the control algorithm.