Research On Knowledge Discovery In Scheme Design Of Large-Span Steel Space Structures

People get together ample examples used for reference in structural design and fabrication, along with rapid development of large–span steel space structures. On the other hand, development of modern structural engineering science enables information contained in these examples to be applied efficiently, and that just solve a engineer problem of large–span steel space structures scheme design being correspondingly indigent. Based on this, aiming at the problem of obtaining large–span steel space structures scheme design knowledge, using rough set to be the base of calculation approaches, this paper researches its application in structural intelligence scheme design knowledge discovery system. The main contents as follows:1. The core process of knowledge discovery-basic knowledge of data mining is expounded detailedly. The advantage of rough set theory is explained by comparison with other approaches. The model of factor system and model of structural system of form selection of large–span steel space structures suiting rough set theory is constructed by analysis.2. The basic knowledge of rough is introduced, and the application process of which is confirmed in data mining. The detailed process of using standard rough set theory to carry through form selection of large–span steel space structures data mining is explained by computing material examples. Standard rough set theory depending on original data is put forward.3. In order to solve in a certain extent depending of original data in form Selection of large–span steel space structures, this paper analyses probabilistic rough set theory; researches the approach of applying information entropy to carry through the process of reduction of attributes; applying Bayes decision in picking–up rules, compares probabilistic rough set with standard rough set.4. Correlative argumentation is carried through aiming at the instance of imperfection of information fallen across possibly in form selection of large–span structures, and disposal approach aiming at valued-tolerance relation as attributes absent badly and data supplement approach based on random set as attributes absent not badly is put forward...