Papers by Keyword: Regression Analysis

Abstract: Based on the physical models of PROSPECT and SAIL, hyperspectral data of different scenes were simulated. According to the simulated data, we built 7 spectral indexes highly linear correlated to vegetation canopy water content, and analyzed the relationship between spectral indexes and canopy water content. Then we built a multiple linear model of canopy water content with the spectral indexes that are highly correlated with FMC (Fuel Moisture Content). At last, using the model, the vegetation canopy water content from Hyperion was obtained. It shows that the spectral indexes: II, NDWI, Ratio1200, Ratio975, WI1, are highly correlated with canopy water content, and the multiple linear model built with them produces an effective result.

Abstract: A critical challenge in prediction of material property is the accuracy of estimation for regression coefficient between the structure or process of material and its macroscopic property. One source of the estimation errors is measurement errors which commonly exist in practice. To provide guidance on the use of simple linear regression methods in measurement error modeling for prediction of material property, we investigated and compared least squares (LS) and orthogonal regression (OR) theoretically. And their applications in prediction of tensile strength for quenched and tempered steel 45 were presented as an example. OR has better performance than LS in the prediction of material property in presence of measurement errors under certain conditions.

Abstract: Machining simulation can quick supply valuable data to help improving machine processing quality and efficiency. And it is playing an important part in the development of modern machining theory. In this paper, the drilling process of an engine block was simulated in the software of DEFORM-3D. The geometry models of cutter and workpiece were simplified to avoid too low efficiency of simulation running. The mesh weight factors were specified to keep geometry models from changing too much from the real situation. 11 simulation experiments were carried out. And through regression analysis of the result data, the drilling load formulas were obtained to predict the axial force and moment in the drilling process of engine block.

Abstract: This article using the method of orthogonal test analysis the law of honeycomb ring part in aero-engineer surface discharge mark diameter by EDG. In the impulse voltage under certain conditions,by changing the peak current and pulse width in the interelectrode study both of the impact of the largest diameter of GH3536 surface discharge mark, and build a mathematical model of the regression equation, derive relatively accurate empirical formula. Provide a good reference and guidance for the actual production of the EDG.

Abstract: This paper establishes correlation analysis model to calculate the profile of trees and the distribution of leaves, and estimate mass of leaves. The establishment of the proposed model makes an in-depth analysis of sample data and determines the interrelationship of all relevant factor which make it easy to operate and applicable to all types of trees. If there are sufficient types of trees, more factors can be considered to make the model more accurate.

Abstract: In order to reduce the elastic recovery of a sheet material and eliminate a great number of solid dies used in the forming process of various shapes, a flexible stretch forming process (FSFP) is considered in this study. Especially, the relationship among design variables, such as the punch size, objective radius of curvature, and elastic pad thickness is quantitatively evaluated to find out their respective influences on the shape errors of a formed sheet plate using the statistical method based on the FE simulation result planned by the three-way factorial design. The shape errors are divided into two types based on the material behavior according to the widthwise- and tensile- directions. The correlations of the shape errors and the design variables are estimated through the Pearson correlation analyses. The punch size has a strong positive linear correlation with the widthwise- and tensile- shape errors, and the correlation between the objective curvature radius and tensile-direction shape error is weak and negative. Although the effect of the elastic pad thickness is less than those of the other variables, it prevents effectively surface defects. Subsequently, the mathematical model is assumed to clarify their relationship. Two regression equations are estimated in terms of the design variables regarding the widthwise- and tensile- shape errors. The shape errors could be inferred by the assumed model in the particular combination of the design variables; then, the acceptable punch size and elastic pad thickness can be determined according to the objective curvature radius.

Abstract: In order to make the reed leaf fiber soft, increase its separating degree and remove impurities, the method of pretreatment in removing impurity was studied. The orthogonal method was used to optimize the treatment condition and the optimal conditions were obtained at a concentration of sodium hydroxide of 18g/L, treating temperature of 100°C and holding time of 90min. It also revealed that temperature holding time had the greatest influence on the degumming effect of the fibers. Under these conditions, the content of pectin and hemicellulose could reach about 2.6% and the content of cellulose could come up to 92.8%. Furthermore, the binary quadratic orthogonal regressive experiment was used to optimize the degumming process. The longitudinal directions of the fibers were examined by SEM before and after the processing.

Abstract: Based on measured data of Wuxinzhai tunnel in expressway of Shanyin to Pinglu, mathematical methods were used to make regressing analysis of measured date on crown settlement and periphery convergence. According to the time-effect of soil mass excavation in New Austrian Tunneling Method, the relations between rock deformation and time during the tunnel construction was analyzed. The results of analysis could provide an theoretical basis for modifying design and construction , and also could offer an example for similar projects in this area.

Abstract: This paper proposes the back-propagation neural network (BPN) and applies it to estimate the slump of high-performance concrete (HPC). It is known that HPC is a highly complex material whose behaviour is difficult to model, especially for slump. To estimate the slump, it is a nonlinear function of the content of all concrete ingredients, including cement, fly ash, blast furnace slag, water, superplasticizer, and coarse and fine aggregate. Therefore, slump estimation is set as a function of the content of these seven concrete ingredients and additional four important ratios. The results show that BPN obtains a more accurate mathematical equation through learning procedures which outperforms the traditional multiple linear regression analysis (RA), with lower estimating errors for predicting the HPC slump.

Abstract: Using the industrial limestone, fly ash and pure chemical reagents as raw materials, the blast furnace slag was prepared in the simulation condition of the actual slag-making process. Using uniform design method, the influence of blast furnace slag composition factors such as quaternary alkalinity, ratio of CaO/MgO and ratio of SiO₂/Al₂O₃ on the glass content of blast furnace slag were studied in the present in investigation, the relationship between glass content and the various factors has been obtained through regression analysis, and the main influence factors and the optimum blast furnace slag compositions been found out. The results showed that, the impact of various factors on the glass content of granulated blast furnace slag was quaternary alkalinity > ratio of SiO₂/Al₂O₃ > ratio of CaO/MgO. According to the significance test of the regression equation and the verification experiment, it was concluded that the optimum compositions of granulated blast furnace slag were: the quaternary alkalinity was 0.85, the ratios of CaO/MgO and SiO₂/Al₂O₃ were 11.5 and 6.5 respectively, and in this condition, the glass content of granulated blast furnace slag reached to 98.47%.