Papers by Author: Bai Lin Liu

Paper TitlePage

Authors: Bai Lin Liu, Hui Yun Zou, Xi Chen
Abstract: In order to solve low accuracy, human effects and complexity in elevator safety management evaluation, a method based on machine learning was proposed. The method adopts safety checklist to collect data of elevator safety related conditions, comprehensively considering the importance and influence of every factor, which influences the safety on the basis of the safety checklist analysis and fuzzy set. To complete the process of the risk assessment and evaluation, we use machine learning combined with maintenance knowledge of evaluation, which provide users with comprehensive and effective corrective measures and suggestions. Applications show that the method can find potential leak of elevator system management.
Authors: Bai Lin Liu, Lei Li
Abstract: In order to improve the effectiveness of fault diagnosis for mechatronic system, new integrated testing instrument was developed. The testing instrument was integrated with experience, detected data and complex technical principles. The structure of the integrated testing instrument was introduced. The system is divided into fault diagnosis expert system and signal processing device. Fault diagnosis expert system software is to complete the human-computer interaction, testing process control, test result analysis processing, output display and fault diagnosis. Signal processing instrument includes test signal acquisition, signal conditioning, data acquisition, and data communication. Experiments show that the instrument can find the fault efficiently and improve the maintenance efficiency of a certain type of mechatronic system.
Authors: Bai Lin Liu, Wen Chen, Lei Li, Qin Ren Xiong
Abstract: To improve the performance data management efficiency of welded pipe used in the 2nd west to east gas pipeline, a distributed performance data acquiring and management system for welded pipe was developed. The system was established by using two-level three-tier Client/Server model. Performance data for welded pipe from different factories was acquired at clients and input to local database. The data then can be analyzed at local, or be transferred to the server through networks for unified storage and analysis. The System has functions as data management, file transferring, standard technology condition management, statistical analysis, figure displaying and statistical analysis report generation, etc. The system was implemented using VC++. Oracle and Access is adopted as database for server and client respectively, XML is as the data encapsulation for transferring file. The developed system plays a significant role in analyzing and evaluating the whole quality of welded pipe, and is useful for pipeline quality control.
Authors: Bai Lin Liu, Li Xing Gao
Abstract: To solve the problem that large training samples and slow speed in diagnosing based on support vector classifier, a hybrid classification algorithm applying attribute reduction of rough set and classification principles of SVM to diagnose diseases was proposed. RS was used to preprocess the attributes on condition that no effective information was lost. Redundant attributes and conflicting objects from decision table was deleted. And the dimension and complexity in the process of SVM classification was reduced. Then it classifies and forecasts objects by SVM classifier, so as to achieve the diagnosis for case. Experiments show that it can improve the rate of diagnosis under the reasonable reducing accurate rating after RS reducing information, and can improve the speed of diagnosing diseases when SVM dealing with much disease information.
Showing 1 to 4 of 4 Paper Titles