Research On Network Fault Diagnosis For Mechanical Parts Of Numerical Controlled Machine

The numerical controlled machine is obtained more and more widespread application as a mechatronics and technology-intensive product. Its technical level and capacity have become the important symbol to judge a national industry modernization level and to manifest a national comprehensive technical strength. As the production's key equipment, the fault in any part will cause working accuracy to reduce, even machine trip out, halt in production. These faults will not only creates the huge economic loss but also endanger to the personal safety in serious. At the same time, as a result of its technological advance, structure complex, moreover the numerical control system model is too many and renews quickly. All those had created the serious lack of servicemen. Therefore, it is an important question which should be solve urgently that use the existing expert experiences to guarantee numerical controlled machine security, reliability and capability, to raise numerical controlled machine 's service efficiency and enterprise's economic benefit. This topic has studied a neural network and expert system integrative network failure diagnosis system. It is based on the Web and in view of numerical controlled machine's mechanical breakdown. It can carry on the common mechanical breakdown diagnosis which occurs on numerical controlled machine.The paper has carried on the diagnosis system's design and the development according to the actual need of the numerical controlled machine mechanical fault, the Microsoft Corporation's ASP and MathWorks Corporation's MATLAB are used as develop platform. A fault diagnosis system has established which integrated the expert system and the neural network. In neural network's structural design process, one dynamic structure optimization method has been used to determination the node number of concealment level, it can determine the reasonable effective network architecture more convenient. "The mechanical device condition examination and the diagnosis virtual instrument system" the topic's earlier period research results has been used to gathering, processing the signal, the time domain and the frequency range analysis has been done to obtain the needing parameter, then the training and the test of this neural network has completed.The neural network and the expert system both of them are as the independent module in this system to carry on the failure diagnosis. The user is permitted to choice sole module or the two unions' carries on the diagnosis according to the actual need. The network diagnosis system may realize resources sharing fully about the expert experience, the technical knowledge and so on, reduces the time of failure diagnosis and service. The system is confirmed by the failure knowledge and sampled data. The result shows that this system has provided the advanced practical to the numerical controlled machine's mechanical fault diagnoses. The man-machine contact surface of this system is friendly, the operator is simplicity. It raised the numerical controlled machine's mechanical fault diagnosis level and efficiency. The modular principle is used to the system design; every part of this system can carry on the diagnosis independent non-interference, moreovercaused system's expansion to be very convenient.