顾及不同约束准则的变形并联组合预测模型研究  

STUDY ON DEFORMATION PARALLEL COMBINATION PREDICTION UNDER DIFFERENT RESTRICTION CRITERION

作　　者：高彩云[1,2] 崔希民[1] 高宁[2] 

机构地区：[1]中国矿业大学(北京)地球科学与测绘工程学院,北京100083 [2]河南城建学院测绘工程学院,平顶山467044

出　　处：《大地测量与地球动力学》2014年第3期91-94,共4页Journal of Geodesy and Geodynamics

基　　金：国家自然科学基金项目(41071328)

摘　　要：根据权重求取的不同约束准则,对变形并联组合预测模型的构建进行研究。分别构建了非最优、最优权、灰色综合关联度定权、熵权、神经网络非线性5类变形并联组合预测模型,讨论了5类约束准则下的权重求取特点,并比较了预测的效果。分析表明,基于神经网络的非线性组合预测具有良好的特性和较高的预测精度。The parallel combination deformation prediction models were studied. According to the weighted constraint criterion, five kinds of deformation parallel combination prediction models were established, including sub- optimal weight combination, optimal weigh combination, synthesis relational grade of grey-weight, entropy-weight and artificial neural network-weight. Based on the characteristics of the models, the effects of parallel combination prediction under different constraint criterion are compared. The results of analysis with real data show that the nonlinear combination prediction based on neural network is the best one.

关 键 词：变形 并联模式 约束准则 权值 组合预测 

分 类 号：P258[天文地球—测绘科学与技术]