Wireless MEMS sensor network based intelligent diagnosis system for manufacturing system

This paper aims to develop a remote monitoring and diagnosis architecture for a manufacturing system, by creating a more miniature and intelligent wireless system. These concepts are essential when developing a flexible remote monitoring and diagnosis architecture for a manufacturing system. A wireless MEMS (Micro-Electro-Mechanical Systems) sensor network based intelligent diagnosis system is proposed and implemented. Test results verify the feasibility of the architecture. The concept of 5-Layers for remote monitoring and diagnosis architecture is introduced, including (1) Signal acquisition and processing layer. (2) Data transmission layer. (3) Feature extraction layer. (4) Performance assessment layer. (5) System integration layer. Based on the concept of 5-Layers for remote monitoring and diagnosis architecture, a WMSN (wireless MEMS sensor network) based intelligent performance prediction and fault diagnosis system for the CNC spindle is proposed and implemented: the different rotating speeds of the spindle motor produce vibrating signals that are acquired by MEMS sensors. Features are extracted using the wavelet packet method. The data is then fused using an envelopment surface method in combination with an ANN (Artificial Neural Network). In order to assess the condition of the spindle motor, the features extracted from the acquired signals are input into the training model of the ANN, so output serves as a performance index. The output data can be sent to the SQL server to integrate the remote monitoring and diagnosis systems of the manufacturing system. A CNC spindle motor commonly featured in manufacturing systems is chosen to verify this architecture. Under the normal and eccentric spindle states, system modeling is built up for a high speed (3,000RPM), medium speed (2,000RPM) and low speed (1,000RPM) The test results of samples of different rotating speeds under the normal and eccentric spindle states show the effect of the intelligent diagnosis system and verify the feasibility of the architecture in terms of miniaturization, wirelessization and intelligentization.