Research On Remote Intelligent Fault-diagnosis Of CNC Machine Tools Based On BN

In modern manufacturing, the various and complex requirement of industry makes the CNC machine tools more and more automatic and networking. While a remote monitoring and maintenance system with the module of intelligent fault -diagnosis is the basic and indispensable unit for automatic and networking machine tools. This can not only increase the efficiency of productivity, but also the degree of automation and flexibility. Therefore, research concerning the fault-diagnosis and remote monitoring for machine tools is significant and meaningful.The thesis is based on and supported by the project of the Science and Technology Project of Zhejiang Province (No. 2006C11067). On the basis of analyzing the advantages of Bayesian Networks when researching problems of uncertainty, a concrete model about the quality of workpiece of CNC lathe was established. Besides, a remote monitoring system was designed as well and the Bayesian model established before was embedded in eventually.The establishment of Bayesian Network model in this thesis replies on the analysis of traditional fault net trees. Firstly, determine the set and domain of variables. Then, construct the structure of Bayesian Network. Finally, find the conditional probability table for each variable in the model. According to the study around the theory of Bayesian Network, the particular procedure of establishing a Bayesian network is elaborated concerning a specific research target-the quality of CNC lathe's workpiece. The model designed is structural unambiguously and calculative briefly as well as understanding easily. The choices of nodes in model can be increased and decreased according to the real conditions and experimental environment, and its experience probabilities can be adjusted based on samples and expert's knowledge. With the accumulation of fault samples, the precision of reasoning in the model will be more and more accurately. This is referentially meaningful to the research of fault-diagnosis about CNC machine tools.Considering the established Bayesian networks and experimental conditions, the thesis designed the rules of feature extractions and obtained features for reasoning by measuring the vibration signals of three points of CNC lathe.