Remote Monitoring And Fault Diagnostic System For RP Machine Based SOM Neurol Network

As an important part in modern advanced manufacturing technology and system, remote-monitoring and trouble-shooting have been attached more importance on. At the same time, many new theories have been developed about monitoring and diagnosis. Remote monitoring technology for modern machine is a novel approach combined with computer technology,communication technology,aritificial intelligence and trouble-shooting technology.At first, the methods and development of monitoring is described in this thesis. After that, disadvantage of traditional monitoring methods is analyzed. Instead of traditional monitoring and diagnostic methods,a framework based SOM neural network and Java 3D is introduced. Then, the difficulty and key technology in this process is listed.Furthermore, the machining framework, characteristic of FDM (Fused Deposition Modeling) Rapid Prototype Machine and the core controller-PMAC, which selected by the monitoring system, is analyzed. Combined with the data monitoring system needed,a data collecting method based PMAC is introduced. The real-time data is collected and saved by serial communication,dll and database technology.Then, after analyzing the fault characteristic of FDM,a diagnostic modeling based SOM neural network is introduced.The SOM neural network model is trained and sorted through the fault eigenvectors,which are results of fuzzy mapping the fault symptom of FDM.The SOM neural network model diagnoses the machining process of FDM through sorting the real-time FDM data.Finally,through setting up three-layer web structure, the project of remote monitoring anddiagnose is tested.Moreover, the effect, which the SOM-based remote monitoring technologymake on the web-based manufacturing environment, and its foreground of application is analyzed.