Water quantity prediction is the base and premise of water price calculating . this paper uses moving tendency forecasting modeling , gm forecasting modeling and bp neural forecasting modeling to forecast the water requirement of the future , evaluates the forecasting results , and confirms the forecasting results ; the industry water price elasticity index and the resident water price elasticity are calculated with the multi - linear regression method ; the water resources value is evaluated with the marginal opportunity cost method considering the transferring water , other parameters are evaluated and estimated by using some methods of connecting with objective laws and estimation 用水量预测是水价制定的前提和基础，本文在进行水量预测时，采用移动平均法、灰色预测法和bp神经网络进行预测，并对预测结果进行了综合评价，确定出合理的预测结果；采用多元线性回归方法确定工业用水价格弹性和居民生活用水价格弹性指数；采用跨流域调水情况下的边际机会成本方法确定当地的水资源价值；采用主观判断和客观规律相结合的方法对其它一些参数进行了确定。
First , strategy analyzing : through analyzing the inner and outer environment factors such as history , resources , competence and its strengths and weaknesses , we identified the firm ' s competitive advantages , core competence and long term goal . especially , we compare and arrange the order of civil listing cement enterprises in china through establishing a series of relevant indexes and fuzzy subsets method . we forecast the firm ' s next 5 years manufacture capability by recession analysis 论文分析了企业的内外环境，归纳出企业优劣势及企业发展的机会和方向，特别通过对企业历史、资源、能力的分析，总结出企业的战略目标及核心能力，通过建立相关指标体系及模糊聚类对水泥行业上市公司的竞争地位进行了比较分析与排序，通过二元回归方法对秦岭水泥的生产规模进行预测。
Because the supply water system is large - scale , the pump stations are many , the ways of supply water are complicated , and the adjustable spare is large , the whole data in the optimization of supply water , the flux and its variety regulations in the control stations ca n ' t be acquired . the paper predicts supply water during the process based on the theory of artificial neural network control . the optimization has two steps 在实际供水系统中很难准确地获取整个管网的所有基础数据，主要难以预知所有节点的逐时流量及其变化规律，因此成为供水系统的优化调度的难点，本文采用人工神经网络对供水量进行预测，应用二级优化调度，第一级以整个供水系统的费用最小为目标，用回归方法求出各供水节点的供水量；第二级利用流量跟踪的方法对每个供水节点(泵站)实现优化运行，以期在泵站内达到最大节能目的。
Quantitative structure - activity relationship is a method building a statistical model . the model can quantificationally predict structure - activity relationship of molecule , and bioactivity of new molecules can be known . structure of molecule , which is described by parameters in physical chemistry , biochemistry and quantum chemistry , includes functional group , minor structure , molecular fragment , chemical composition 本文提出了一种定量构效关系的处理方法,即应用支持向量回归方法解决定量构效关系,并且应用这种方法预测了芳香烃化合物生物降解度与喜树碱化合物抗肿瘤生物活性,取得了比较令人满意的结果。
Abstract : some of the present models of forecasting zipingpu ' s flood can t fully discover the information in flood process and others are impractical . so a staistical method , projection pursuit regression is applied in forecasting peak flood and flood process . comparing with other models , the new one gets high precision and is easy to be operated in practice 文摘:针对现有紫坪埔洪水预报模型或不能充分挖掘样本信息或不便于实际应用的问题，现用投影寻踪回归方法建立了紫坪埔洪水预报模型，分别对洪峰和洪水过程进行了预报，并与其他方法进行了对比，取得了满意的效果，可以作为紫坪埔洪水预报的新方案。
The variable bandwidth local linear regression method we used in this paper have the advantages of both the local linear regression method and the variable bandwidth idea . the variable bandwidth henced the flexibility of the estimation . and it make this method can fit the spacial complex curve very well . and the asymptotic results of the estimation found the theoretic base for find the best variable bandwidth and drive the pratical best variable bandwidth from data directly 本文所用的变窗宽局部线性回归方法，继承了局部线性回归的优点，并且使用变窗宽提高了所得估计的可塑性。并使之能成功地处理空间非齐次曲线等复杂形状的曲线拟合问题。所得估计的渐近结果为求渐近最优窗宽方案以及直接从数据估计最优变窗宽提供了理论基础。
This work researched and discussed the modeling and optimization methods and their applications in chemical process . the main contributions of this work are as follow : 1 ) discussed the partial least square regression ( plsr ) and multi - cyclic subspace regression ( mcsr ) , investigated radial basis function neural network ( rbfnn ) , analysed genetic algorithm ( ga ) and its weakness 本文正是对化工过程建模、优化及其应用的一种研究与探讨，主要包括以下工作： 1 )论述了偏最小二乘回归方法( plsr )与多元的循环子空间回归方法( mcsr ) ，讨论了径向基函数神经网络( rbfnn ) ，分析了遗传算法( ga )及其应用中的不足。
( 6 ) the reservoir operation function is established utilizing the optimal dispatching results . the state variable and decision variable of the operation function is discussed , and the stepwise regression method is used to derive the operation function . because of the nonlinear features of the function have n ' t been reflected in traditional regression methods , the back - propagation neural network model is introduced to establish the operation function ( 6 )利用水库优化调度结果建立水库调度函数，在分析水库调度函数各特征量的基础上，介绍了用逐步回归方法建立水库调度函数的具体过程，考虑到传统回归方法未能反映调度函数的非线性特性，引入bp神经网络模型求解模型，建立水库调度函数。
