Application of Slicing in Component-based System Regression Testing

In component-based system regression testing, the dependence model of component-based system should be built in order to confirm the part which will be affected by modify. Since high reusability and high complexity, the existing dependence models may not be applicable to component-based systems, a hierarchical dependence model is proposed to describe component-based. Through the forward slicing of modified points in the model, the modified and affected parts can be gotten, then traverse these parts and get affected testing path, thereby regression testing examples can be chosen effectively, the efficiency of regression testing can be improred.     

Regression Applied in Continuous Mathematical Forecast

Data Mining can distill the connotative, unknown and potential value information or pattern from many data. It is a problem that must be settled during the developing of Decision Support System how to discover knowledge from Data Warehouse. In order to realize the continuous mathematical forecast that will be used in the Marketing Decision Support System for Industrial Enterprise, the authors use the method of regression to discuss how to design the algorithm for Data Mining, how to design the model for Data Mining, and so on. When the algorithm applied to the sales forecast, the result processing to simulation data indicates it can realize the expected effect.     

THE DEVELOPMENT OF COMPUTER AIDED NORM OF WORKING HOUR PLANNING SYSTEM

Based on Group Technology, part families of the norm of working hour are formed according to the similarity among components. With the progressive regerssion, the models of the norm of working hour are built for every part family. A computer aided norm model building program and a computer aided norm of working hour planning system are developed. The system is successfully integrated with CAD/CAPP system. Some examples of the model building and calculation results are illustrated as well.     

Transmission Congestion Management Based on Power Influence Coefficient

Basing eliminating on the idea that the generator who has the most influence on the power flow of line is mostly in favor of transmission congestion, the model of congestion management with the object of minimizing adjustments of bufions varied with generation outputs is proposed. It can eliminate the congestion effectively by judging the generators who should be adjusted and the amounts of adjustment. The test on IEEE-RTS24 buses system and other practical systems show that the proposed method is effective.     

The Convergence Rates of Improveed Kernel Estimators for Regression Function

The kernel function of kernel estimator for regression function is often defined with compact support, the moment of Y is bigger than one. The convergence rate of the improved kernel estimator is discussed under the improved kernel function. The result is also discussed under random censorship. The class of applicable kernels include those having unbounded support and even not integral .The condition of moment of Y is discarded under the improved kernel estimator. The results are wholly the same as usual and generalize the relative results of the paper .     

The Algorithm of Searching out the Least and Most Loops by Making Use of Direction Factor

Based upon the conception of the least loop,the most loop and direction factor presented by the author firstly,the algorithm of searching out the least and most loops has been designed.By building the adjoining matrix, confirming the start-point of searching,vectorizing the searching sides,calculating the direction factor of adjoining sides,we can search out the next side of the least loop or the most loop rapidly.With a valid searching side being confirmed at every searching and the searching scope being reduced by forming degenerate graph,the time compexity of the algorithm is much less. In the research of the automatic calculating system of construction quantities, the problem of houses partition and out-wall partition had been solved succesfully by utilizing this searching algorithm.     

基于LS-SVM的小水电站年发电量智能预测模型

针对小水电站年发电量序列的特点,将最小二乘支持向量机(least squares support vector machine,LS-SVM)回归模型引入年发电量预测领域,并给出了相应的过程和算法。与常规基于人工神经网络(artificial neural net-works,ANN)的智能预测方法比较,该模型优点是明显的:①将神经网络迭代学习问题转化为直接求解多元线性方程;②整个训练过程中有且仅有一个全局极值点,确定了预测的稳定性。最后,一个实际的预测例子表明:该模型实现容易、预测准确,适用于小水电站预测。     

高度对色斑法测量雨滴粒径影响的试验研究

以前的色斑法雨滴率定只考虑色斑直径与雨滴真实粒径的关系，忽略了高度对率定的影响。在大量试验的基础上，通过回归分析，得到了以色斑直径和高度为变量的率定公式。研究表明，水滴下落高度小于5m时需考虑高度的影响；高度大于5m时，高度的影响可以忽略。通过相关性分析，公式的拟合性较好，对以后的色宽法测量雨滴粒径有指导意义。     

基于随机森林回归算法的小麦叶片SPAD值遥感估算

使用机器学习中的随机森林(RF)回归算法构建小麦叶片SPAD值遥感反演模型。以2010—2013年江苏地区试验点稻茬小麦3个生育期(拔节、孕穗、开花)的叶片为材料,结合我国自主研发的环境减灾卫星HJ-1对研究区域进行同步监测,分析了各生育期叶片SPAD值与8种植被指数间的相关性;以0.01水平下显著相关的植被指数作为输入参数,使用RF回归算法构建了每个生育期的小麦SPAD反演算法模型,即RF-SPAD模型,以支持向量回归(SVR)和反向传播(BP)神经网络算法构建的SVR-SPAD模型和BP-SPAD模型作为比较模型,以R2和均方根误差(RMSE)为指标,分析了每个生育期3个模型的学习能力和回归预测能力,结果表明:RF-SPAD模型在3个生育期都表现出最强的学习能力,R2和RMSE在拔节期分别为0.89和1.54,孕穗期分别为0.85和1.49,开花期分别为0.80和1.71;RF-SPAD模型在3个生育期的回归预测能力都高于BP-SPAD模型,高于或接近于SVR-SPAD模型,R2和RMSE在拔节期分别为0.55和2.11,孕穗期分别为0.72和2.20,开花期分别为0.60和3.16。     

基于最小二乘支持向量机模型预测田面糙率

田面糙率是影响地面灌溉质量的重要参数。基于最小二乘支持向量机建立了两类4个田面糙率预测模型,并进行了验证。结果表明第一类模型预测值(即作物地采用LSSVM-N-I3、裸地采用LSSVM-N-I1,翻耕地采用LSSVM-N-I2)相对误差最大值为9.7%;第二类模型预测值(即LSSVM-N-II模型)相对误差最大值为10.5%,由此可见两类模型都具有较高的预测精度,可以用于田面糙率的预测。