The Research On Wheel Load Dynamic Detecting Of Weigh-in-Motion Systems

the social consequences of violations of listed companies,large-scale violations of disturbing the securities market,investors lose confidence,even more serious can cause systemic financial risk.In this paper,the research on the early warning of the listed companies is carried out,which mainly adopts the support vector machine and the transfer learning method.This paper's main results are as follows:First to between 2010 and 2014 of the 1535 a-share listed companies as the research sample data,integrated with Fisher score and MRMR two feature selection algorithms,screening and companies have the feature of maximum correlation properties,then use support vector machine(SVM)to establish early warning model.Through the comparative analysis found that the early warning model of feature attribute screening for higher precision.Results show that a proportion of the first largest shareholder,net asset per share,long-term loans to total assets ratio,tobin Q value and current debt ratio,number of shareholders' general meeting,industry type affected by 20 property violations to the company.Second,about 1535 samples was divided into two major categories of manufacturing industry and the manufacturing industry,is proposed based on Fisher distinguish subspace in a banach space transformation of manufacturing industry to the manufacturing industry migration method.This method will be the difference between the source domain and target domain Bregman distance is adopted to improve the said,because the source domain and target domain in the form of feature mapping,by studying the source domain and target domain category information,training and iterative classifier in the feature space.The empirical results show that using the industry data model for the industry analysis of the early warning,warning effect than cross-industry model precision is higher.After building a subspace by principal component analysis algorithm,the manufacturing industry company data modeling,predict the rate of manufacturing companies.By reducing the difference between domains,improve the quality of the model,improve the cross-industry early-warning model accuracy.