Research On Real-time Monitoring And Prognostic Technique Of CNC Machine Tool Based On Internet Of Things

With the rapid development of machinery manufacturing industry, the CNC systems become more complex and the corresponding price becomes higher. The possible breakdowns of the expensive CNC machine tools would bring the CNC processing enterprises huge econo mic losses. The machine tool manufacturers also need to monitor the sold CNC machine tools remotely to provide the users more comprehensive after-sales services. Therefore, the condition monitoring and fault diagnosis of CNC machine tool ha s been paid special attention by the machine tool manufacturers and CNC processing enterprises. In order to meet the demand of the equipment health management, this paper will focus on the real-time monitoring and prognostic technique of CNC machine tool and build a kind of system based on this technique. The main works are undertaken mainly in following aspects.1. Research on multi-channel En Dat2.2 interface. In order to obtain absolute position value and fault information of machine axis, the multi-channel bidirectional digital interface technique based on En Dat2.2 protocol is studied. The hardware and software system of the multi- channel En Dat2.2 interface are built by the System-on-a-Programmable-Chip(So PC) technique. The En Dat2.2 port logic based on the Avalon bus and the So PC system software are designed by the analysis of reading and writing time constraints of the En Dat2.2 protocol conversion IP core and reading and writing process specification of the Avalon bus slave port.2. Research on wireless sensor network of machine tool. In view of the application of wireless communication technique in the real-time monitoring and prognostic system of CNC machine tool, the wireless sensor network of machine tool is studied. According to the specific requirements of the machine tool monitoring, the kind of radio frequency chip in the wireless sensor network of machine tool is selected, and the hardware of the data acquisition node and the sink node is designed. The wireless ad hoc network protocol of machine tool based on the handshake communication and survival time is designed. The network protocols of the acquisition node and the sink node are achieved by the protocol data packet format. Through the wireless ad hoc network protocol of machine tool, the wireless sensor network of machine tool can incorporate the network-wide active acquisition nodes into the sensor network and delete the acquisition nodes which do not respond within a period of time.3. Research on the multi-thread communication mechanism in the Internet of things of mechine tool. In order to solve the problems from the traditional multi- thread mechanism in the Internet of things of mechine tool, the synchronization and communication mechanism between threads in the Internet of things environment is studied. Through the comparison of the several common thread synchronization tools, the mailbox is chosen to solve the thread synchronization problems in the Internet of things of mechine tool. Aiming at the problems that the traditional scheduling algorithms cannot give priority to the urgent mail in the full mailbox and cannot replace the low real- time normal mail, the scheduling algorithm based on urgent mails and the LRU algorithm based on normal mails are purposed. The contrast experiments showed that the above two algorithms can meet the real-time requirement of the real-time monitoring and prognostic system of CNC machine tool.4. Research on the temperature feature extraction algorithms based on linear regression. According to the change of machine tool equipment temperature caused by gradual faults and sudden faults, the corresponding linear regression modes are established. The linear feature extraction algorithms based on gradual faults and sudden faults are proposed on the basis of the linear regression modes, and the least square estimate for the next moment temperature value is achieved. The simulation experiment shows that the temperature feature extraction algorithms based on linear regression can not only achieve the on- line monitoring and prognostic of gradual faults and sudden faults, but also eliminate the interference of the "burr" phenomenon for the feature extraction results effectively.5. Research on the vibration feature extraction algorithm based on wavelet packet decomposition. According to the theory of wavelet transform, the vibration feature extraction algorithm is proposed based on wavelet packet decomposition. The information in the frequency domain of the vibration signal is decomposed into the different sub frequency bands of the wavelet packet decomposition tree. The energy feature vector of the vibration signal in a time window can be extracted from the energy statistics of the corresponding coefficient sequences in the sub frequency bands. Because the vibration energy distribution of the rotating machinery has a great similarity between the same faults and a great difference between the different faults, the online diagnosis of the rotating machinery’s fault types can be achieved by calculating the degree of similarity between the different energy feature vectors.