Font Size: a A A

Research On Data Mining Methods Of Bamboo Wood Properties

Posted on:2019-11-03Degree:MasterType:Thesis
Country:ChinaCandidate:X X YueFull Text:PDF
GTID:2428330551959470Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Bamboo wood is an important forest resource.The study of its properties is the basis for the rational use of bamboo wood resources.The traditional research on bamboo wood properties is based on expert experience and lacks specific data support.These factors lead to the low level of information of bamboo wood data and the unclear key factors affecting the quality of bamboo.This has further affected the screening of excellent wood materials and the comprehensive utilization of bamboo materials.Based on the data mining technology,this paper builds a database of bamboo germplasm basic properties and studies the optimization of key factors and bamboo classification methods for excellent bamboo wood.And developed a data mining system for bamboo wood properties,to provide data bases and scientific methods for the rational and effective use of bamboo wood.The specific research content is as follows:(1)The selection method of key factors with properties feature was proposed based on the mechanical properties of ReliefF.First,using the three feature selection methods of ReliefF,CFS and Wrapper to optimize the 21 attribute parameters of bamboo material properties.Then,the screened features were combined with support vector machines to construct mechanical regression prediction models for flexural strength,shear strength in straight lines,compressive strength in straight lines and tensile strength in straight lines,and then verified and analyzed.The results show that the support vector regression model constructed based on the feature variables selected by ReliefF has higher prediction accuracy and is obviously superior to CFS and Wrapper methods.It is proved that the ReliefF method is feasible for the selection of the key material factors.This method clarifies the key factors for the mechanical properties of the four bamboo materials.(2)Bamboo classification model based on PCA-IAGNES was constructed.Principal component analysis(PCA)was used to reduce the dimensionality of 25 wood characteristics,and the first six principal components combined with AGNES algorithm were selected for classification modeling.The model parameters were optimized by the distance threshold.A classification model based on PCA-IAGNES for the selection of bamboo materials with excellent wood properties was constructed,and data examples were verified on the bamboo database.The results show that the PCA-IAGNES can gather the bamboo materials into 3 categories and the classification results are basically consistent with the relevant national standards.Therefore,data mining methods can be used to classify bamboo materials and screen out excellent bamboo materials.(3)Bamboo wood properties data mining system was developed.Using the MVC framework to integrate Java,Weka and MySQL databases on the Eclipse platform,a bamboo wood properties data mining system was designed and developed.The system functionally designed the user rights management module,data basic operation module,data mining module,database maintenance and search engine module,and system monitoring module.The method proposed was implemented in the paper.This research has been innovative in the selection of key factors for the mechanical properties of bamboo materials based on ReliefF and the bamboo classification method based on PCA-IAGNES.It provides scientific basis and research methods for the clear identification of key materials of excellent materials and bamboo classification,and has important practical significance for the rational development and utilization of bamboo resources.
Keywords/Search Tags:bamboo wood, properties database, data mining, bamboo classification, key factors
PDF Full Text Request
Related items