Network Topology Information Mining Algorithm Design And The Application Research

The research on the rules of influences of various factors on fatigue strength and fatigue life is crucial. In preliminary design stage of the parts, the rules could be used to estimate the fatigue strength and fatigue life directly. Meanwhile, it cannot be ignored that the existence of mutual interference among the fatigue influencing factors. For this reason, it is a very meaningful work to research on fatigue strength influencing factors and the interactions between them. In view of the similarities between researches on fatigue strength’s influence factors and bioinformatics analysis of multi-factors for tumor, our research start from the identification of differential expressed biological network (pathway). Then present a new algorithm for network topology information mining. Finally, the new method is applied to the mechanical fatigue strength’s influence factors analysis. The main work in the dissertation can be summarized as follows:1) Proposed a directed edge in a network modeling methods based on Markov chain model to appraisal the performance. In the methods, the genetic information flow or signal transduction in pathway is mathematically depicted as a state transition probability matrix; the activity of interactions between genes is measured.2) Evaluated the overall performance score of network by the score of edges in it to identify the sub-network associated with the sample phenotype. This process helps us explore the potential link between overall network information and sample phenotype. Firstly, the average score of edges is defined as the overall performance of network. Then a permutation is used to obtain the statistically significance.3) Applying this new network topology information mining algorithm to analyze which factors of parts significantly associated with the fatigue strength and fatigue life. We build a complex network in which the node represents the potential factor, and the edge is the interaction between two factors. Then performance score of each edge is calculated by our method and the best edge is found. Finally, we decide which factor has a significantly influence for the fatigue strength and fatigue life.